[{"data":1,"prerenderedAt":6748},["ShallowReactive",2],{"$fwuenr1OZcDw8TjcC-BzsU1lhSGBFolBhwdGCSmwMXKw":3,"$ftoUUvay03wjsVvLRB4QkJLEjxwDkVF3M27glSEGre1w":4},null,{"2026-08-08":5,"2026-08-09":3460},[6,36,64,90,118,164,189,213,234,260,279,303,323,347,379,404,429,450,473,496,519,541,565,584,601,622,641,660,681,706,728,746,766,787,808,827,849,877,896,919,942,964,983,1004,1024,1045,1060,1090,1109,1129,1151,1172,1193,1212,1233,1250,1275,1302,1323,1343,1362,1384,1403,1424,1443,1466,1486,1514,1533,1559,1578,1599,1618,1639,1659,1683,1704,1725,1746,1767,1789,1806,1821,1842,1861,1882,1903,1922,1943,1964,1985,2004,2025,2045,2071,2090,2109,2128,2149,2170,2193,2208,2227,2247,2265,2285,2301,2320,2341,2361,2382,2402,2423,2442,2463,2483,2504,2525,2546,2562,2583,2602,2623,2648,2667,2686,2703,2723,2738,2758,2777,2798,2826,2845,2863,2882,2901,2934,2953,2974,2993,3016,3035,3054,3080,3099,3114,3133,3152,3171,3184,3210,3229,3244,3264,3286,3305,3325,3344,3364,3379,3398,3419,3439],{"id":7,"room":8,"start":11,"end":12,"language":13,"track":14,"speakers":19,"zh":27,"en":31,"tags":34,"uri":35},"39SXDY",{"en":9,"zh-hant":10},"TR409-2","","2026-08-08T12:30:00+08:00","2026-08-08T13:00:00+08:00","中国語",{"id":15,"name":16},541,{"en":17,"zh-hant":18},"Taiwan L10N within FLOSS projects","臺灣自由軟體在地化社群",[20],{"id":21,"avatar":22,"zh":23,"en":26},"FWQNJ3","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FFWQNJ3_iRhBvkh.webp",{"name":24,"bio":25},"Kisaragi Hiu","KDE zh-TW translator, amateur coder. Interested in a million things but not good enough at most of them.",{"name":24,"bio":25},{"title":28,"describe":29,"type":30},"提供 KDE 台文翻譯的嘗試","台語正在經歷振興以利台語文化不至於失傳。如同其他閩南語系語言，台語大部分話者皆未以文字書寫台文，以至於書寫台文以其母語者的規模而言極為罕有。要改善這個情況的其中一部是讓軟體的使用者介面提供台文版本。作為達成使用者介面台文化的一步，我開始進行了將 KDE 軟體譯為台語的作業，並希望與自由軟體和台語振興社群分享目前為止的歷程。","Talk",{"title":32,"describe":33,"type":30},"Translating KDE into Taigi","Taigi is a language in a process of being revitalized so that it will survive into the next generations. Taigi, just as other Hokkien languages, has historically not been used in writing by most speakers. To bring written Taigi to this era and to provide options for people who already speak the language, it is necessary that user interfaces are translated into this language. As a stepping stone, I have embarked on an effort to provide Taigi translations for KDE projects, and wish to share the journey so far with the wider community.",[],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002F39SXDY",{"id":37,"room":38,"start":40,"end":41,"language":42,"track":43,"speakers":48,"zh":56,"en":60,"tags":61,"uri":63},"3WHJYA",{"en":39,"zh-hant":10},"TR511","2026-08-08T15:40:00+08:00","2026-08-08T16:30:00+08:00","Mandarin",{"id":44,"name":45},544,{"en":46,"zh-hant":47},"Anonynet Community (anoni.net)","匿名網路社群 anoni.net",[49],{"id":50,"avatar":51,"zh":52,"en":55},"UYKEPE","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FUYKEPE_Y9UGubd.webp",{"name":53,"bio":54},"Liangcc","CC went on a soul search for what's missing in the blockchain space and continues to get lost. He is interested in humanity and economics research. He is also distracted by the future of software development and cryptography proofs.\r\nPreviously, he built zero knowledge applications in the Ethereum ecosystem.\r\n\r\nCC 一直在尋找區塊鏈領域中缺少的東西，然後持續迷失其中。他對人文與經濟學研究感興趣，同時也分心於軟體開發的未來與密碼學證明。\r\n過去，他在以太坊生態系統中開發零知識應用程式。",{"name":53,"bio":54},{"title":57,"describe":58,"type":59},"隱私支付實作工作坊：從龍捲風現金到隱私池","本工作坊從使用者視角出發，介紹龍捲風現金（Tornado Cash）的設計哲學與隱私池（Privacy Pool）的改進。我們會利用道具與實機操作，展示兩樣工具的使用方式。重點放在實際操作與使用上的隱私眉角，而非密碼學細節。\r\n時間允許下，我們也會介紹這些工具的沿革與在現實世界的衝擊。","workshop\u002Fpanel",{"title":57,"describe":58,"type":59},[62],"Elementary","https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002F3WHJYA",{"id":65,"room":66,"start":68,"end":69,"language":70,"track":71,"speakers":75,"zh":83,"en":86,"tags":87,"uri":89},"3XWBLQ",{"en":67,"zh-hant":10},"TR412-1","2026-08-08T11:20:00+08:00","2026-08-08T11:50:00+08:00","English",{"id":72,"name":73},540,{"en":74,"zh-hant":74},"Python Track by Taipei.py",[76],{"id":77,"avatar":78,"zh":79,"en":82},"787FUL","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002F787FUL_wY0skgp.webp",{"name":80,"bio":81},"Petr Andreev","Specializes in CPython internals, optimization, and high-performance computing.  \r\nDriven by GPU acceleration, CPU vectorization. Evolved from ML systems to CPython core research engineer.  \r\n8+ years leading teams in AI, maths, and physics. PyCon speaker.  \r\nLecturer at Moscow Institute of Physics and Technology – top 1 Russian university.  \r\nOpen to talks and collaboration.",{"name":80,"bio":81},{"title":84,"describe":85,"type":30},"CPython 3.15: Updates Rivals Will Use — Profiler, JIT, noGIL, Lazy Imports, UTF-8","CPython 3.15 brings updates that change how we debug, start, and scale Python systems — not just syntax sugar.\r\n\r\nIn this talk, I’ll __go deep__ on 2 high-impact topics:\r\n- Tachyon profiling: find bottlenecks with low overhead\r\n- FreeThreading updates vs. Go\u002FJava\u002F.NET, where CPython wins, where it doesn’t\r\n\r\nThen I’ll do a __fast sweep__ of:\r\n- explicit lazy imports: faster startup for your app\r\n- frozendict, and its design\r\n- unpacking in comprehensions: your eyes love it!\r\n- UTF-8 by default: already feeling more confident?",{"title":84,"describe":85,"type":30},[88],"Intermediate","https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002F3XWBLQ",{"id":91,"room":92,"start":94,"end":95,"language":96,"track":97,"speakers":102,"zh":110,"en":113,"tags":116,"uri":117},"737RTQ",{"en":93,"zh-hant":10},"TR412-2","2026-08-08T14:30:00+08:00","2026-08-08T15:00:00+08:00","中文",{"id":98,"name":99},542,{"en":100,"zh-hant":101},"Taiwan JVM Team","JVM 台灣代表隊",[103],{"id":104,"avatar":105,"zh":106,"en":109},"TCPLT9","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002F%E5%A4%A7%E9%A0%AD1_3bXlygs.jpg",{"name":107,"bio":108},"朱尚禮 samzhu","Java 開發者",{"name":107,"bio":108},{"title":111,"describe":112,"type":30},"用 Spring AI Agent Client 開發 CLI Agent 任務編排工具","如果你和我一樣同時訂閱了 Claude Code、Gemini、Codex, 不同任務想用不同的 Agent CLI,但每次切換就要手動搬運專案上下文、複製程式碼、重新解釋需求, 工作流支離破碎。訂閱費繳得不少,實際使用體驗卻像在三個分身之間當客服。\r\n後來發現了 Spring AI Agent Client 這個 Spring AI Community 在 2025 才釋出的新專案，可以在 Java 端統一編排多家 AI Agent 的方案，它把 Claude Code、Gemini、Codex、Amazon Q 等 CLI Agent 抽象成同一套介面,就像當年 ChatClient 統一了各家 LLM API 一樣。\r\n如果你也有類似困擾可以參考看看",{"title":114,"describe":115,"type":30},"Building a CLI Agent Task Orchestrator with Spring AI Agent Client","If you're like me and have subscriptions to Claude Code, Gemini, and Codex all at once, you've probably hit the same pain point: different tasks call for different Agent CLIs, but every time you switch, you end up manually shuffling project context around, copy-pasting code, and re-explaining requirements. Your workflow falls apart. You're paying for three subscriptions, yet it feels like you're playing customer-support agent for three different clones of yourself.\r\nThen I came across Spring AI Agent Client, a new project released by the Spring AI Community in 2025. It lets you orchestrate multiple AI Agents from the Java side under a unified interface — abstracting Claude Code, Gemini, Codex, Amazon Q, and other CLI Agents behind a single API, much like how ChatClient unified the various LLM APIs back in the day.\r\nIf you've been running into the same kind of problem, this talk might be worth a look.",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002F737RTQ",{"id":119,"room":120,"start":122,"end":123,"language":96,"track":124,"speakers":129,"zh":156,"en":159,"tags":162,"uri":163},"77RHET",{"en":121,"zh-hant":10},"TR411","2026-08-08T09:30:00+08:00","2026-08-08T10:20:00+08:00",{"id":125,"name":126},521,{"en":127,"zh-hant":128},"AI Open Governance","AI開放治理",[130,137,144,151],{"id":131,"avatar":132,"zh":133,"en":136},"WUXBDZ","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FWUXBDZ_CqrQdHP.webp",{"name":134,"bio":135},"數位發展部","數位發展部在臺灣的人工智慧發展中，對內從算力、資料、人才、行銷媒合及資金等五大面向著手，鼓勵公私協力，並透過風險分類框架及先進評測環境，打造安全可信賴的生態環境，對外積極與美國國家標準暨技術研究院（National Institute of Standards and Technology，NIST）和歐盟《人工智慧法》（EU AI Act）等標準接軌，期望將臺灣的人工智慧產業推向國際。",{"name":134,"bio":135},{"id":138,"avatar":139,"zh":140,"en":143},"ATEXDF","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FATEXDF_FH16ifL.webp",{"name":141,"bio":142},"郭銘傑","郭銘傑是國立臺灣大學政治學系專任副教授兼社會科學院國際長，同時也是台灣網路治理論壇常務理事與多方利害關係人。\r\n郭教授在美國加州大學聖地牙哥分校獲得政治學博士。\r\n他的研究專長包括國際政治經濟學、政治心理學、  民主韌性、 公眾外交、全球網路治理等。\r\n郭教授目前正在進行的研究之一是探索世界各國人工智慧政策實踐的多樣性，並建立數位主權在全球人工智慧價值鏈中興起的社會科學分析架構。\r\n\r\nJason Kuo is an Associate Professor in the Department of Political Science and Associate Dean for International Affairs at the College of Social Sciences, National Taiwan University.\r\nHe is also an Executive Board Member and Chair of the Multistakeholder Steering Group of the Taiwan Internet Governance Forum.\r\nProfessor Kuo received his Ph.D. in Political Science from the University of California, San Diego.\r\nHis research interests include international political economy, political psychology, democratic resilience, public diplomacy, and global internet governance.\r\nOne of Professor Kuo’s current research projects explores the diversity of artificial intelligence policy practices across countries and develops a social science analytical framework for understanding the rise of digital sovereignty in the global artificial intelligence value chain.",{"name":141,"bio":142},{"id":145,"avatar":146,"zh":147,"en":150},"ECRSQW","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FECRSQW_91h8Sti.webp",{"name":148,"bio":149},"Teemo","臺大醫院資訊工程師，長期專注生成式 AI、流程自動化與知識管理整合，具備將先進技術落地至企業與醫療場域的深厚實務經驗。曾多次入選總統盃黑客松與國家級 AI 競賽，並受邀於多家醫院與產業單位授課，是推動非技術人員掌握 AI 工具效率的專業講師。",{"name":148,"bio":149},{"id":152,"avatar":3,"zh":153,"en":155},"MVC7NR",{"name":154,"bio":10},"湯家碩",{"name":154,"bio":10},{"title":157,"describe":158,"type":59},"人工智慧與開源的對話：國際趨勢與政策發展","這些年，各國政府都對人工智慧（ Artificial Intelligence，AI ）做出包含隱私保護、智慧財產權等諸多原則性指引，但實務上，多數國家都尚未對 Vibe coding 程式碼錯誤、代理型AI（AI Agent）指令錯誤、演算法偏誤、AI幻覺等問題的責任歸屬做好準備。\r\n\r\nAI 因人而生，以程式碼為基礎，產出程式碼，為人服務，最終也應服膺於人與程式碼。那在這個嶄新的時代，各國政府是如何看待 AI 與人的關係，以及組成 AI 和 AI 產生的程式碼，在這場講座，我們將邀請各方專家學者為我們講述，也彼此辯論。",{"title":160,"describe":161,"type":59},"AI x Open Source: Global Trends & Policy Landscape","Over the past few years, governments worldwide have laid out foundational guidelines for Artificial Intelligence, focusing on critical pillars like data privacy and intellectual property. Yet, in practice, a massive regulatory gap remains. Most jurisdictions are still caught off-guard when it comes to assigning liability for the \"Vibe Coding\" era—whether it’s flawed code, erratic AI Agent commands, algorithmic bias, or the persistent challenge of AI hallucinations.\r\n\r\nAI is built by humans, grounded in code, and designed to generate even more code to serve humanity. Ultimately, it must remain accountable to both its human creators and the underlying logic of its syntax. In this brave new era, how do governments perceive the evolving relationship between humans and AI, or the interplay between the code that builds AI and the code AI produces?\r\n\r\nIn this session, we’ve gathered a panel of leading experts and scholars to unpack these complexities, share their insights, and engage in a spirited debate on the future of our digital governance.",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002F77RHET",{"id":165,"room":166,"start":11,"end":12,"language":96,"track":168,"speakers":173,"zh":181,"en":184,"tags":187,"uri":188},"7Q8NTC",{"en":167,"zh-hant":10},"TR212",{"id":169,"name":170},532,{"en":171,"zh-hant":172},"Open LLM Tech: Core Technologies","Open LLM Tech: 開源模型技術",[174],{"id":175,"avatar":176,"zh":177,"en":180},"UDMCG7","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002Fprofile6_PaR8OZx.png",{"name":178,"bio":179},"Martin Chang","Systems software engineer working on HPC, GPGPU, and AI.",{"name":178,"bio":179},{"title":182,"describe":183,"type":30},"可解釋性與應用：人類的最終防線","LLM 與相關技術為科技與人類帶來了巨大的轉變。但我們依然不知道 LLM 內部的原理與如何確保 LLM 不會有惡意行為。可解釋性技術提供了少數的切入點。與其存外部在訓練過程中控制語言模型的行為，不如直接打開模型，去直接探索甚至控制LLM的行為。\r\n\r\n但這麼重要的技術卻鮮少被討論跟應用。這裡我們打開他的面紗，為未來控制更強大的 AI 爭取希望\r\n\r\nCode: https:\u002F\u002Fgithub.com\u002Fmarty1885\u002Fllama.cpp\u002Ftree\u002Frwkv-edit",{"title":185,"describe":186,"type":30},"Mechanistic interpretability and applications: The final fronteir of hacking","LLM changed the course of tech and humanity for better or for worse. But safety has and still is a problem. No one can guarantee if LLMs are evil or misaligned. Instead of trying to make LLMs safe during training by different means. Mechanistic interpretability provides a different way - try to detect and change LLM behavior by opening them up and see what is going on. \r\n\r\nMechanistic interpretability is an important tech and not an easy one. Yet they are rarely discussed or developed publicly. Let's change that. If not so future us down the road can control much more powerful AI then right now.\r\n\r\nCode: https:\u002F\u002Fgithub.com\u002Fmarty1885\u002Fllama.cpp\u002Ftree\u002Frwkv-edit",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002F7Q8NTC",{"id":190,"room":191,"start":193,"end":194,"language":96,"track":195,"speakers":199,"zh":207,"en":210,"tags":211,"uri":212},"7VR89U",{"en":192,"zh-hant":10},"TR213","2026-08-08T12:10:00+08:00","2026-08-08T12:40:00+08:00",{"id":196,"name":197},534,{"en":198,"zh-hant":198},"System Software",[200],{"id":201,"avatar":202,"zh":203,"en":206},"DUDBPL","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FDUDBPL_9OoQ0EV.png",{"name":204,"bio":205},"vic","I'm Chill Vic, a software engineer who enjoys sharing my findings and knowledge through social media.",{"name":204,"bio":205},{"title":208,"describe":209,"type":30},"Beyond epoll: Rethinking High-Performance Network Runtime with io_uring","epoll has been the foundation of modern high-performance Linux servers for years, powering event-driven systems such as Nginx, Redis, and many async runtimes. However, as workloads continue scaling, the readiness-based model of epoll introduces increasing complexity around event loops, syscall overhead, and asynchronous file I\u002FO.\r\n\r\nThis talk introduces why Linux developed io_uring, how it differs from epoll, and what architectural changes it brings to modern async runtimes. We will briefly compare epoll and io_uring, explore how Rust async runtimes can integrate with io_uring through the Minato WebSocket server project, and discuss why io_uring matters not only for networking but also for file I\u002FO and database systems such as PostgreSQL.\r\n\r\nThe session focuses on practical system software design, runtime architecture, and the future direction of Linux asynchronous I\u002FO.",{"title":208,"describe":209,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002F7VR89U",{"id":214,"room":215,"start":216,"end":217,"language":42,"track":218,"speakers":220,"zh":228,"en":231,"tags":232,"uri":233},"7VSVUT",{"en":9,"zh-hant":10},"2026-08-08T10:00:00+08:00","2026-08-08T10:30:00+08:00",{"id":15,"name":219},{"en":17,"zh-hant":18},[221],{"id":222,"avatar":223,"zh":224,"en":227},"FKUXQS","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FFKUXQS_Oamvcoh.webp",{"name":225,"bio":226},"黃敬群","網上慣用 \"jserv\" 代稱，任教於國立成功大學，專注作業系統、編譯器，和虛擬機器等領域。曾任聯發科技、台達電子、鴻海科技集團，和工業技術研究院，和國家太空中心的技術顧問，並長期投入開放原始碼軟體開發。",{"name":225,"bio":226},{"title":229,"describe":230,"type":30},"用 zhtw-mcp 搭配 AI 寫出更像台灣人的中文 | Write Taiwanese Mandarin: zhtw-mcp as a Linguistic Linter for Traditional Chinese","大型語言模型產生的繁體中文，常混入中國用語、過度簡化的漢語、錯誤標點與不精準術語，這些輸出會在文件、教學、開放原始碼專案與技術文章中逐漸污染書寫習慣。zhtw-mcp 是個透過 Model Context Protocol (MCP) 整合到 AI 助理的繁體中文書寫檢查器，可依中華民國教育部標點、國字標準字體與兩岸詞彙差異，檢查並修正 AI 產生的中文。\r\n\r\n本議程介紹為什麼 AI 中文輸出容易被不精準來源混雜影響，如何將中文潤稿轉化為可測試、可重現、可融入進開發流程的工程議題，並展示 zhtw-mcp 如何自動改善台灣繁體中文書寫材料的品質和文化傳承。\r\n\r\nLLMs often generate Traditional Chinese that looks fluent but mixes zh-CN vocabulary, Simplified-Chinese phrasing, inconsistent punctuation, and unstable technical terminology. zhtw-mcp turns this problem into an engineering workflow: a Model Context Protocol server that checks, fixes, and gates zh-TW text using Taiwan-specific language rules.\r\n\r\nThis talk presents the design of zhtw-mcp, the challenges of handling ambiguous terminology and false friends, and how linguistic linting can be integrated into AI-assisted writing, documentation, teaching material, and open-source development workflows.",{"title":229,"describe":230,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002F7VSVUT",{"id":235,"room":236,"start":238,"end":239,"language":96,"track":240,"speakers":244,"zh":252,"en":255,"tags":258,"uri":259},"7VWJXN",{"en":237,"zh-hant":10},"TR514","2026-08-08T10:10:00+08:00","2026-08-08T10:40:00+08:00",{"id":241,"name":242},516,{"en":243,"zh-hant":243},"Ruby Taiwan",[245],{"id":246,"avatar":247,"zh":248,"en":251},"X99AKT","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FX99AKT_uZy234u.webp",{"name":249,"bio":250},"Lilith","Ruby 工程師，目前在5xruby開發團隊工作。日常工作橫跨後端 API、前端畫面渲染，對開發有趣又實用的工具這件事有偏執的興趣。",{"name":249,"bio":250},{"title":253,"describe":254,"type":30},"網站改版了，然後你的 Spider 就壞了：Ruby 爬蟲維護的現實","維護一個爬蟲不難，但維護超過 100 個、針對不同金融機構網站的爬蟲，是完全不同等級的挑戰。                                \r\n                  \r\n  這場分享來自真實的 production 系統——一個持續從日本各大基金公司抓取 PDF 月報的 Rails   \r\n  專案。當目標網站從 REST API 悄悄遷移到 Nuxt.js SPA、當基金的列表頁因為        \r\n  ReactVirtualized 只渲染可見列導致找不到目標基金、當PDF版月報突然只剩 HTML       \r\n  版的月報——你的 spider 就這樣靜靜壞掉了。                                            \r\n                                                                                      \r\n  我會分享三種最常見的「壞法」與對應的 Ruby 解法、如何用 Module \u002F Concern 組織大量      \r\n  spider 的共用邏輯，以及最重要的工程判斷：什麼時候該修，什麼時候該直接刪掉。",{"title":256,"describe":257,"type":30},"Your Scraper Just Broke — Again: Hard-Won Lessons from 100+ Ruby Spiders in Production","Maintaining one web scraper is easy. Maintaining over 100, each targeting a different\r\n  financial institution's website, is a completely different beast.                     \r\n   \r\n  This talk is drawn from a real production Rails system that continuously downloads    \r\n  monthly report PDFs from Japanese asset management companies. When DWS silently\r\n  migrated their REST API to a Nuxt.js SPA, when JPMorgan's ReactVirtualized table only \r\n  rendered 12 visible rows and your target fund wasn't one of them, when Capital Group\r\n  stopped offering PDF files and switched to HTML-only reports — your spiders just    \r\n  quietly broke.\r\n\r\n  I'll walk through the three most common failure patterns and their Ruby solutions, how\r\n   Module and Concern help organize shared logic across a large spider fleet, and the\r\n  most important engineering judgment of all: when to fix, and when to just delete.",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002F7VWJXN",{"id":261,"room":262,"start":122,"end":216,"language":96,"track":264,"speakers":266,"zh":273,"en":276,"tags":277,"uri":278},"8CZ7ER",{"en":263,"zh-hant":10},"TR510",{"id":44,"name":265},{"en":46,"zh-hant":47},[267],{"id":268,"avatar":269,"zh":270,"en":272},"QWBSWP","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FQWBSWP_otR9Hf6.svg",{"name":47,"bio":271},"我們是關注「匿名網路」、「網路自由」的一群人，透過 Tor、Tails、OONI 工具喚起一般大眾對網路自由的重視，並在臺灣推動隱私保護的實踐。\r\n\r\n專案起初聚焦於 OONI，透過調查臺灣 ASNs 在 OONI 資料庫中觀測資料的完整性。後來我們逐步擴展定位，進一步專注在「匿名網路」議題的推廣與工具建立。2025\u002F08，我們在國立台灣科技大學順利舉辦了「匿名網路工作坊」，招募到第一批社群夥伴，一同展開 2026 年的工作。\r\n\r\n2026 年，社群聚焦三大主題：\r\n\r\n- 個人隱私指引：整理可實際操作的隱私保護指引，依情境（日常、敏感工作、高風險）提供工具與步驟。\r\n- Tor Relay 校園建立：與 EFF、Tor 合作推動校園中繼節點，已有台師大成功案例。\r\n- 匿名支付：探索現金以外情境下的匿名支付（如穩定幣、區塊鏈應用），含法規與實作面向。\r\n\r\n[anoni.net](https:\u002F\u002Fanoni.net\u002F)",{"name":47,"bio":271},{"title":274,"describe":275,"type":30},"匿名網路社群 anoni.net 介紹：開源匿名工具、社群實踐與 2026 三大主題","匿名網路社群 anoni.net 是台灣在地的開源社群，這一年持續推廣 Tor、Tails、OONI 這些自由開源的匿名網路工具。\r\n\r\n這場開場先說明社群的緣起與運作方式，以及社群如何以開源協作參與國際專案，把 Tor、Tails、OONI 與 Tor University Challenge 的介面翻成正體中文、回報問題、協助在校園架設中繼節點。接著從記者、公民團體、人權倡議者實際會遇到的處境切入：通訊可能被攔截、身分可能被追蹤，消息來源的安全也連帶受影響。這裡也說明為什麼這類高風險工作特別倚賴原始碼可被公開檢視的開源工具，封閉的黑箱軟體無從驗證它有沒有被動過手腳。最後介紹社群 2026 年推進的三個主題，依情境分級的個人隱私指引、校園 Tor Relay 中繼節點建立，以及在合法前提下對匿名支付的開源工具探索。",{"title":274,"describe":275,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002F8CZ7ER",{"id":280,"room":281,"start":283,"end":284,"language":96,"track":285,"speakers":289,"zh":297,"en":300,"tags":301,"uri":302},"8F8QRJ",{"en":282,"zh-hant":10},"RB102","2026-08-08T10:45:00+08:00","2026-08-08T11:15:00+08:00",{"id":286,"name":287},539,{"en":288,"zh-hant":288},"PostgreSQL Taiwan",[290],{"id":291,"avatar":292,"zh":293,"en":296},"Z8FQ8V","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002F257742836_304947081666893_8527341263200811560_n_h0qIGBU.jpg",{"name":294,"bio":295},"Peter","喜歡專注在後端程式、資料庫與軟體系統架構等領域，使用 PHP、Python與JavaScript作為軟體系統開發已經超過 9 年的時間，同時對於開放源始碼抱持著高度的熱情，有時遇到開源專案有 bug 時，就會嘗試動手來協助修正。截至目前為止，已經有超過 1000 個 PR 被 merge，也在貢獻開源專案的過程中，學到許多撰寫程式碼的技巧，同時也不吝在各大技術研討會進行技術的分享，如：COSCUP、MOPCON以及PyConTW等。",{"name":294,"bio":295},{"title":298,"describe":299,"type":30},"How to integrate Python, Grovvy, PostgreSQL and NiFi to complete hardware factors collection?","時至今日，為了要讓資料流程能夠更自動化與更容易的進行管理，許多的軟體系統對於資料處理面流程中，陸續導入了資料中台、資料網格與資料經緯等技術，試圖讓資料流、資料管理等議題更加地容易。\r\n\r\n在COSCUP 2025時，講者分享了如何使用PostgreSQL搭配NiFi實現簡易自動化資料收集的良好經驗。本議程中，基於上述的議程，講者將會分享實際的應用專案案例，如何使用NiFi、PostgreSQL、Python與Groovy進行互相搭配、結合與寫程式的方式，打造出屬於自己各個項目的硬體指標收集與彙算的資料流，並分享相關資料流、架構設計與建置的經驗。",{"title":298,"describe":299,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002F8F8QRJ",{"id":304,"room":305,"start":11,"end":306,"language":70,"track":307,"speakers":309,"zh":317,"en":320,"tags":321,"uri":322},"8RZ3ZV",{"en":93,"zh-hant":10},"2026-08-08T12:45:00+08:00",{"id":98,"name":308},{"en":100,"zh-hant":101},[310],{"id":311,"avatar":312,"zh":313,"en":316},"TSTEFD","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FTSTEFD_TfSA4Vr.webp",{"name":314,"bio":315},"Sunyoung Shin","DevOps Engineer",{"name":314,"bio":315},{"title":318,"describe":319,"type":30},"From Ruby to Java: Migrating a Notification Platform to Microservices","At Remember (a career networking platform in Korea), our notification service had been running on a legacy Ruby-based architecture for years.\r\n\r\nAs the number of users and notification channels grew, we started facing challenges in scalability, maintainability, and operational complexity. The system had accumulated technical debt, making it increasingly difficult to introduce new features and ensure reliable delivery.\r\n\r\nInstead of performing a risky big-bang migration, we gradually rebuilt the notification platform as Java-based microservices while continuing to serve production traffic.\r\n\r\nIn this talk, I will share:\r\n- Why we decided to migrate from Ruby to Java\r\n- How we decomposed the existing architecture\r\n- Strategies for running old and new systems in parallel\r\n- Challenges around message delivery, consistency, and monitoring\r\n- Lessons learned from operating the migration in production\r\n\r\nThis session focuses on practical engineering decisions rather than theory, providing insights that teams can apply when modernizing long-running systems.",{"title":318,"describe":319,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002F8RZ3ZV",{"id":324,"room":325,"start":327,"end":193,"language":328,"track":329,"speakers":333,"zh":341,"en":344,"tags":345,"uri":346},"8UMS73",{"en":326,"zh-hant":10},"AU","2026-08-08T11:40:00+08:00","英文",{"id":330,"name":331},545,{"en":332,"zh-hant":332},"UbuCon Asia",[334],{"id":335,"avatar":336,"zh":337,"en":340},"AL9MFC","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FAL9MFC_YMaDqMs.webp",{"name":338,"bio":339},"Alfred Neumayer","Music connoisseur & hobby tinkerer, Hip Hopper, comedy fan, spicy eater.",{"name":338,"bio":339},{"title":342,"describe":343,"type":30},"Ubuntu Touch Progress Update 2026","This is a regular status update of what has landed and what's landing in Ubuntu Touch.\r\n\r\nWe'd like to show you what we have been working on over the last year. From upgrades to our UI toolkit to Qt6 over to the quest of bringing Mir 2 to supported devices, right next to how we land those in a storage-limited environment. And of course we have Snap updates to show.\r\n\r\nMany pieces to pick up and you'd surely like to know why, what and what's going on.\r\n\r\nMore info: https:\u002F\u002Fevents.canonical.com\u002Fevent\u002F146\u002Fcontributions\u002F953\u002F",{"title":342,"describe":343,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002F8UMS73",{"id":348,"room":349,"start":68,"end":69,"language":42,"track":351,"speakers":356,"zh":371,"en":374,"tags":377,"uri":378},"98GWFJ",{"en":350,"zh-hant":10},"TR513",{"id":352,"name":353},530,{"en":354,"zh-hant":355},"Cyber Resilience","網路韌性",[357,364],{"id":358,"avatar":359,"zh":360,"en":363},"E7TTJV","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FE7TTJV_zMiCTWl.webp",{"name":361,"bio":362},"Johnson","Cofacts 專案發起人與核心開發者。長期致力於透過開源技術與群眾協作，對抗 LINE 上的不實訊息。近年專注於探索 LLM 應用落地與系統架構開發，致力於降低群眾查核門檻，期盼藉由科技陪伴公民社會一同面對資訊操弄的挑戰。\r\n\r\nI founded the Cofacts project in 2016 and serve as its core developer. Cofacts is an  open-source civic tech project dedicated to fighting disinformation on closed messaging apps.\r\n\r\nOver the past few years, my work has focused on integrating emerging technologies, such as LLMs and semantic search algorithms, into our system. I also host weekly open meetings to maintain project transparency and engage with the community, ensuring our open-source journey is accessible to everyone interested in civic tech.",{"name":361,"bio":362},{"id":365,"avatar":366,"zh":367,"en":370},"GE77WE","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FGE77WE_j6QLbbB.webp",{"name":368,"bio":369},"Benson Su","Cofacts 貢獻者、邊緣人，原本是打(算)滾(出) 遊戲業的 Unity 工程師，2018 年剛好手上專案比較閒，就在工作時間研究 web 前後端、發了一個 PR 進 Cofacts，然後就越做越多。自詡是 Cofacts 的「清潔工 程師」，一路都在修 bug、做小功能，做著做著反而變成把在 Cofacts 學到的東西拿回工作上用。\r\n\r\n想要告訴大家一開始點錯技能樹不用懊悔，只要加入開源社群就好（X",{"name":368,"bio":369},{"title":372,"describe":373,"type":30},"抵禦資訊操弄的開源實踐：Cofacts 群眾查核平台的經驗與挑戰","民主社會仰賴公民的集體智慧，然而這份信任基礎正遭到資訊操弄的侵蝕。\r\n\r\n「Cofacts 真的假的」自 2016 年起建立開源的群眾查核共筆平台，讓在意假訊息的民眾撰寫查核回應，直接與在 LINE 上收到謠言的群眾對話。\r\n\r\n這十年間，假訊息的處理一直都極具挑戰性：同一則謠言常以文字、截圖、影像等形式不斷變形重現；而近年生成式 AI 的普及，更讓「純 AI 生成影像」的查核與比對難上加難。另一方面，我們在線下工作坊觀察到「群眾防禦」的困境：民眾雖對假訊息感到焦慮，但查核門檻過高，志工常因「怕自己查錯」而陷入自我懷疑與消耗。\r\n\r\n面對不對稱的資訊戰，AI 既是挑戰也是機會。這場分享會將帶過社群查核者會遇到的痛點，以及 Cofacts 打造開源技術解方的經驗。我們也會探討如何讓 AI 扮演「輔助」而非「取代」的角色，在不失去人類的自主性（human agency）的前提下，引領公民跨越查核門檻，一同參與這場抵禦資訊操弄的運動。",{"title":375,"describe":376,"type":30},"Open-Source Practices Against Information Manipulation: Experiences and Challenges of the Cofacts Crowd-Sourced Fact-Checking Platform","Democracy relies on citizens' collective intelligence, a foundation now severely undermined by information manipulation.\r\n\r\nSince 2016, Cofacts has built an open-source, wiki-like collaborative fact-checking platform. It empowers citizens who care about disinformation to write responses, directly communicating with users who receive hoaxes in closed messaging apps.\r\n\r\nOver the past decade, handling disinformation has been inherently challenging: the same hoax often mutates across text, screenshots, and videos. The recent popularization of generative AI has made the verification exceptionally difficult. Furthermore, our offline workshops highlight a critical bottleneck in crowd-sourced defense: despite widespread anxiety over disinformation, the barrier to entry remains dauntingly high. The fear of making mistakes often paralyzes volunteers, leading to severe self-doubt and rapid burnout.\r\n\r\nIn the face of asymmetric information warfare, AI is both a challenge and an opportunity. This session will walk through the pain points experienced by community fact-checkers and Cofacts' journey in building open-source technological solutions. We will explore how to position AI as an \"assistant\" rather than a \"replacement\"—ensuring that human agency is preserved while guiding citizens to overcome verification barriers and collectively join the movement to defend against information manipulation.",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002F98GWFJ",{"id":380,"room":381,"start":12,"end":382,"language":96,"track":383,"speakers":387,"zh":395,"en":398,"tags":401,"uri":403},"997SN8",{"en":39,"zh-hant":10},"2026-08-08T13:30:00+08:00",{"id":384,"name":385},520,{"en":386,"zh-hant":10},"Blockchain and Distributed Ledger",[388],{"id":389,"avatar":390,"zh":391,"en":394},"KBPWBX","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FKBPWBX_LTloAgV.webp",{"name":392,"bio":393},"Ya-wen Jeng","Vivian Jeng 目前任職於 Ethereum Foundation 的 Privacy Stewards of Ethereum 團隊，專注於 ZK（零知識證明）與其他隱私技術的應用推廣。她曾主導並參與開發 Mopro（Mobile Prover）與 Unirep（Universal Reputation）兩項專案，皆為協助開發者更輕鬆將隱私協定整合進自身應用的軟體工具。此外，她也積極參與以太坊社群，包含國際性的 DevCon 和 DevConnect 開發者大會，以及本地的 ETH Taipei 等活動。",{"name":392,"bio":393},{"title":396,"describe":397,"type":30},"零知識證明與自然人憑證身份驗證","在數位身份驗證的時代，如何在不洩漏任何個人資訊的前提下，向服務提供者證明「我是台灣公民」？本議程將介紹我們如何運用零知識證明（ZKP）技術，結合自然人憑證，打造一套隱私保護的公民身份驗證系統。\r\n\r\n自然人憑證（ROC Citizen Digital Certificate）內建 RSA 非對稱金鑰，可由晶片卡對任意訊息進行數位簽章。我們的系統以此為基礎，透過 OpenAC 電路框架設計了一套 ZKP proving scheme，能夠解析並驗證自然人憑證產生的 RSA 簽章，同時在電路層面隱藏所有可識別個人的欄位。\r\n\r\n整套流程完全在使用者裝置上執行：應用程式請求自然人憑證對特定 challenge 進行簽章，隨即於本地端生成零知識證明，最後僅將 ZKP proof 提交至驗證服務。驗證方可確認「此證明由持有合法自然人憑證的台灣公民所產生」，卻無從得知持有人的姓名、身份證字號或任何其他個人資訊。\r\n\r\n為降低開發門檻，我們已釋出完整的 SDK 與範例應用程式，涵蓋從憑證請求、簽章取得、本地 proof 生成到後端驗證的完整流程，讓開發者得以快速整合隱私保護的公民身份驗證功能。\r\n\r\n本議程將涵蓋以下主題：\r\n\r\n- ZKP 與自然人憑證的技術結合原理\r\n- RSA 簽章電路設計與 OpenAC 框架介紹\r\n- On-device proof 生成的效能挑戰與優化策略\r\n- SDK 架構設計與實際整合 Demo 展示\r\n\r\n若你對隱私強化技術（PET）、去中心化身份（DID）、或密碼學應用於公民科技感興趣，這場議程將帶給你從理論到實作的完整視角。",{"title":399,"describe":400,"type":30},"Privacy-Preserving Citizenship Verification Using Zero-Knowledge Proofs and Taiwan's National ID Card","How can a user prove they are a Taiwanese citizen to an online service — without revealing their name, national ID number, or any other personal information? This talk presents our answer: a zero-knowledge proof system built on top of Taiwan's Citizen Digital Certificate (自然人憑證), a government-issued smart card that can produce RSA signatures.\r\n\r\nUsing the OpenAC proving framework, we designed a ZKP circuit that parses and verifies the RSA signature produced by the card, while keeping all personally identifiable fields hidden inside the proof. The entire flow runs on-device: the app requests a signature from the card for a given challenge, generates the zero-knowledge proof locally, and submits only the proof to a verifier. The verifier learns that the prover holds a valid Citizen Digital Certificate — and nothing else.\r\n\r\nTo make this accessible to developers, we have released an open-source SDK and a reference application covering the full pipeline: card interaction, signature acquisition, on-device proof generation, and backend verification.\r\n\r\nIn this session we will cover:\r\n\r\nHow ZKP and the Citizen Digital Certificate work together\r\nRSA signature circuit design and the OpenAC framework\r\nPerformance challenges and optimizations for on-device proof generation\r\nSDK architecture and a live demo\r\nIf you are interested in privacy-enhancing technologies (PET), decentralized identity (DID), or cryptography applied to civic tech, this talk offers a complete view from theory to working implementation.",[402],"Advanced","https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002F997SN8",{"id":405,"room":406,"start":239,"end":408,"language":96,"track":409,"speakers":413,"zh":421,"en":424,"tags":427,"uri":428},"9CYHJT",{"en":407,"zh-hant":10},"TR311","2026-08-08T11:10:00+08:00",{"id":410,"name":411},519,{"en":412,"zh-hant":412},"JSDC X DevFrontier",[414],{"id":415,"avatar":416,"zh":417,"en":420},"MQ7VY3","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FMQ7VY3_FzvvECt.webp",{"name":418,"bio":419},"ShengKai Tang","Software Engineer at Taiwan Mobile",{"name":418,"bio":419},{"title":422,"describe":423,"type":30},"AI 跑得越快，我們反而越容易在螢幕前迷路？","2026 年，AI coding agent 的 context window 突破百萬 token。這意味著模型已經能輕易吞下專案程式碼、理解依賴關係並完成大部分任務。AI 的能力有了巨大的飛躍，分擔了我們許多技術工作。但身為螢幕前的開發者，我們的大腦並沒有跟著升級。當 AI 能一口氣吞吐海量資訊，人類的工作記憶卻依然被困在 7±2 個項目的生物極限。面對龐大專案與快速推進的進度，我們經常在編輯器中看著眾多開啟的分頁，被迫進行頻繁的脈絡切換，陷入「我是誰？我在哪？我現在要幹嘛？」的困境。\r\n\r\n這場分享將以我業餘時間開發的兩個小眾 VS Code 擴充套件為主線。第一款 [Quick Prompt](https:\u002F\u002Fgithub.com\u002Fwinterdrive\u002Fvscode-quick-prompt) 處理了「AI 跑任務 N 時，人已經想到 N+1」的時間差，提供一個內建於開發環境的筆記工具當作思考緩衝區，讓開發者外包記憶、維持思維的連續性。另一款 [VirtualTabs](https:\u002F\u002Fgithub.com\u002Fwinterdrive\u002Fvscode-virtual-tabs) 則是一個用於管理分頁群組的虛擬檔案目錄。在實作完 AI context 匯出以及 MCP (Model Context Protocol) 伺服器等整合功能後，卻發現零星收到的 Issues 幾乎全集中在「標亮當前檔案」、「拖曳排序」、「視窗畫面分組」等幫助人類維持空間感的基礎 UI 需求。\r\n\r\n這場分享將側重於實務面的開發經驗，分享專案中的技術實作，以及從使用者回饋中獲得的哲學反思：在模型能力幾乎能自己解決技術問題的今天，我們開發工具時，解決的究竟是 AI 的瓶頸，還是人類認知負荷的瓶頸？如果你也對開發者工具或 AI 協作體驗感興趣，歡迎參與本場議程。",{"title":425,"describe":426,"type":30},"The Faster AI Runs, The Easier It Is for Us to Get Lost in Front of the Screen?","In 2026, AI coding agents have crossed the million-token context window — they can search entire repos, understand dependencies, and carry out tasks autonomously. The models have improved significantly, taking over a vast amount of our technical workload. However, human working memory is still limited to 7 ± 2 items. While AI processes millions of tokens seamlessly, developers facing dozens of open editor tabs still frequently fall into the trap of \"Who am I? Where am I? What was my next step?\"\r\n\r\nThis talk follows the story of two niche, open-source VS Code extensions I built in my free time. The first, [Quick Prompt](https:\u002F\u002Fgithub.com\u002Fwinterdrive\u002Fvscode-quick-prompt), addresses the asynchronous nature of human-AI collaboration (\"while the AI runs task N, the human has already thought of N+1\") by providing an internal scratchpad to offload memory and maintain cognitive continuity. The second, [VirtualTabs](https:\u002F\u002Fgithub.com\u002Fwinterdrive\u002Fvscode-virtual-tabs), is a virtual file directory for organizing editor groups. After implementing integration features like AI context exporting and an MCP (Model Context Protocol) server, the few issues I received were surprisingly focused on fundamental UI needs—highlighting the active file, drag-and-drop sorting, and window layout management—features that help humans maintain spatial awareness.\r\n\r\nThis session will primarily focus on practical development experiences, sharing the technical implementations from these projects as well as the philosophical reflections inspired by user feedback: In an era where models can solve most technical problems on their own, should our tooling efforts focus on the bottlenecks of AI, or the cognitive bottlenecks of humans? If you are interested in developer tools or human-AI collaboration experiences, you are welcome to join this session.",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002F9CYHJT",{"id":430,"room":431,"start":216,"end":217,"language":96,"track":432,"speakers":434,"zh":442,"en":445,"tags":448,"uri":449},"9E3NRG",{"en":350,"zh-hant":10},{"id":352,"name":433},{"en":354,"zh-hant":355},[435],{"id":436,"avatar":437,"zh":438,"en":441},"SMCCYY","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FSMCCYY_jaythJN.webp",{"name":439,"bio":440},"Irvin Chen","開放文化基金會海纜專案研究員，G0v 韌性松 \u002F Mozilla 貢獻者，摩茲工寮社群空間 keyholder。近期，Irvin 的興趣在理解數位政策與法規，對於網路平台及民主發展的影響，以及試圖深掘「海纜斷光，到底會怎麼樣？」的答案。\r\n\r\nIrvin 同時是前端工程師與開源社群參與者。自 2011 年起擔任 Mozilla 台灣社群聯絡人，負責跨國志工與專案協調，並共同管理摩茲工寮社群空間。他曾在女人迷 (womany.net) 負責網站前端架構規劃，並協助設計 DEI 企業評鑑等策略產品，推動臺灣職場平權與多元包容。Irvin 曾協助將國台語及多個族語納入 Common Voice 開源語音庫，建立釋出第一套免費且自由的台灣原住民多族語音資料集。\r\n\r\nIrvin 關心數位政策與法規對網路及民主發展的影響，今年起亦協助舉辦每月定期「g0v 韌性松」，關注國際海纜斷線對社會帶來的挑戰。\r\n\r\nIrvin Chen is a Submarine Cable Researcher at the Open Culture Foundation (OCF) and an active contributor to the g0v and Mozilla Community. He is also a keyholder of the MozTW Community Space, where he supports community collaboration and open technology initiatives.\r\n\r\nIrvin is both a frontend engineer and an open source community builder. He has served as a Mozilla Taiwan Community Liaison since early '2000, coordinating international volunteers and projects as a member of the Mozilla Reps Council. He was previously responsible for frontend architecture at womany.net, where he contributed to the development of DEI-focused strategic products to promote workplace equality and inclusion.\r\n\r\nHe has contributed to Mozilla Common Voice by helping incorporate Taiwanese languages, including Taiwanese Hokkien and Formosan Indigenous languages, into open speech datasets, helping create the first freely available multi-Indigenous speech corpus of Taiwan.\r\n\r\nIn recent years, Irvin has focused on understanding how digital policies and regulations shape the Internet ecosystem and democratic development. Irvin helps organize the monthly g0v Digital Resilience Hackathon, and is currently focusing on a critical question: what actually happens when submarine cables go dark, the societal impact and broader digital resilience challenges.",{"name":439,"bio":440},{"title":443,"describe":444,"type":30},"海纜斷光到底會怎樣？台灣常用網站的斷網韌性檢測","去年 12 月底，一個海底地震，讓台灣連外的海底電纜壞了七七八八。而受周遭海纜障礙影響，近兩年網路變慢的程度幾乎人人有感！\r\n\r\n假設真的有這麼一天，台灣對外的海纜斷光光了，哪些你我日常使用的網站，還能繼續維持服務？又有哪些你每日需要的網站，屆時會壞光光？\r\n\r\n自 2023 年底的 g0v 第零次韌性松起，作者即持續關心「海纜障礙對於台灣人日常上網會有多大影響」問題。歷經數次大松的 prototype，及多次對外提案，本研究終於獲得 APNIC Foundation 贊助，與 OCF 聯手合作，於 2025\u002F10 開始執行。\r\n\r\n我們擷取了台灣流量前兩千名的網站，對其進行首頁的連線測試。結論是：在對外網路斷光光時，至少有 47% 的網站不會動，另外 43% 也因國際雲依賴而存在高風險。\r\n\r\n立刻查看 resilience.ocf.tw\u002Fweb 檢視測試數據及程式。\r\n\r\n還想知道更多更深入的研究發現，以及我們的建議嗎？歡迎一起加入「海纜斷光，到底會怎樣？」討論吧！",{"title":446,"describe":447,"type":30},"When Submarine Cables Go Dark: Measuring the Resilience of Websites Used in Taiwan","At the end of last December, an undersea earthquake damaged more than half of Taiwan’s communication cables. Combined with ongoing cable incidents in the region, people have already experienced noticeable network slowdowns over the past few months.\r\n\r\nWhat if things get even worse?\r\n\r\nIf Taiwan were to lose all international connectivity, which websites that we rely on every day would still function—and which ones would fail completely? How would undersea cable disruptions affect everyday Internet in Taiwan?\r\n\r\nSince the first “g0v Digital Resilience Hackathon” from late 2023, Irvin has been exploring this central question. This work evolved into a full research project supported by the APNIC Foundation starting in October 2025, in collaboration with the Open Culture Foundation (OCF).\r\n\r\nWe collected and analyzed approximately 2,000 of the most popular websites used by people in Taiwan, performing request-level measurements on their homepage loading behavior. The results show that at least 47% of websites would fail to function properly, plus another 42% exhibit high-risk dependencies, under a complete international disconnection scenario.\r\n\r\nCheck https:\u002F\u002Fresilience.ocf.tw\u002Fweb for code, data, and results.\r\n\r\nWhat do these findings mean for digital resilience—and what can be done about it? Join us to explore the hidden dependencies behind modern web services and discuss how Taiwan can better prepare for a “cables down” scenario.",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002F9E3NRG",{"id":451,"room":452,"start":453,"end":454,"language":96,"track":455,"speakers":457,"zh":465,"en":468,"tags":471,"uri":472},"9FAFNN",{"en":9,"zh-hant":10},"2026-08-08T14:00:00+08:00","2026-08-08T14:50:00+08:00",{"id":15,"name":456},{"en":17,"zh-hant":18},[458],{"id":459,"avatar":460,"zh":461,"en":464},"DHWKZY","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FDHWKZY_I2ST3QI.png",{"name":462,"bio":463},"Sung Chien-chang \u002F Winston Sung","MediaWiki中文暨漢語族社群坑主、MediaWiki原始碼審核員\r\n\r\nWinston Sung於2020年3月10日起開始參與MediaWiki開發，於2023年4月17日正式取得MediaWiki原始碼審核員身分，並於2023年12月29日起擔任MediaWiki中文暨漢語族社群坑主至今。目前致力於推動漢語族語言及臺灣原住民族語言在MediaWiki及維基媒體專案的發展及標準化作業。",{"name":462,"bio":463},{"title":466,"describe":467,"type":59},"MediaWiki暨維基媒體專案臺灣原住民族語言峰會","本議程集合各臺灣原住民語言以及各臺灣本土語言，共同討論在MediaWiki及維基媒體專案下如何強化各語言的發展。",{"title":469,"describe":470,"type":59},"MediaWiki and Wikimedia Projects Formosan Languages Summit","This session is about discussing the development of MediaWiki and Wikimedia projects in Formosan languages and other Taiwan native languages.",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002F9FAFNN",{"id":474,"room":475,"start":476,"end":477,"language":96,"track":478,"speakers":482,"zh":490,"en":493,"tags":494,"uri":495},"9KULK9",{"en":67,"zh-hant":10},"2026-08-08T14:40:00+08:00","2026-08-08T15:10:00+08:00",{"id":479,"name":480},515,{"en":481,"zh-hant":481},"Open-EP (E-Paper) Community",[483],{"id":484,"avatar":485,"zh":486,"en":489},"3WYPUV","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002F3WYPUV_gVKahcC.webp",{"name":487,"bio":488},"蔡丞銳","大家好，我是Triangle，目前是一名熱愛硬體的 Maker。平時喜歡把玩各種感測器與顯示模組，致力於尋找軟硬體整合的最佳解法。\r\n\r\n這次的專案源自我被市面上極度碎片化的電子紙驅動「雷」過無數次的血淚史。秉持著「看不慣就自己造輪子」的開源精神，我決定動手整合出一套跨型號的通用 API，並深入研究底層波形榨出了灰階功能。我深信開源的價值在於為社群填坑、降低開發門檻，希望這次的分享能幫助大家不再受電子紙型號所苦，輕鬆把精力花在打造更酷炫的專案上！\r\n[GitHub](https:\u002F\u002Fgithub.com\u002FmeteorTriangle)",{"name":487,"bio":488},{"title":491,"describe":492,"type":30},"解密 E-paper 驅動：跨型號硬體抽象層設計與灰階 Waveform 實作","電子紙（E-paper）憑藉著超低功耗與護眼特性，在物聯網設備與自造者專案中大受歡迎。然而，開發者往往面臨一個巨大的痛點：「驅動程式極度碎片化」。市面上常見的電子紙模組（如 Waveshare 系列），即使尺寸相同，內部控制晶片與初始化指令也可能完全不同，導致開發者經常需要重複造輪子，難以將程式碼跨型號移植。\r\n\r\n本議程將分享如何打造一套跨型號的電子紙通用 API。透過建立完善的硬體抽象層（HAL），開發者只需撰寫一次程式碼，即可無縫適配多款不同型號的電子紙模組，大幅降低開發與維護成本。\r\n\r\n此外，本議程的另一大技術亮點是「灰階功能的逆向與實作」。許多平價電子紙模組在官方預設下僅支援純黑白顯示，我們將深入探討電子紙的刷新原理，解析如何透過控制硬體波形（Waveform LUT）或演算法，突破硬體原廠限制，在通用 API 的架構下成功榨出平滑的灰階顯示效果。\r\n\r\n這套解決方案目前已實作於 [ESP32 \u002F Arduino \u002F stm32 \u002F Rpi \u002F Rpi pico] 平台，並以 [ C ] 開源。無論你是深受電子紙驅動折磨的 Maker，還是對底層波形控制感興趣的開發者，都能在這場議程中獲得實用的解法與啟發。",{"title":491,"describe":492,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002F9KULK9",{"id":497,"room":498,"start":499,"end":500,"language":96,"track":501,"speakers":503,"zh":511,"en":514,"tags":517,"uri":518},"9RGETL",{"en":39,"zh-hant":10},"2026-08-08T11:30:00+08:00","2026-08-08T12:00:00+08:00",{"id":384,"name":502},{"en":386,"zh-hant":10},[504],{"id":505,"avatar":506,"zh":507,"en":510},"9TAKL8","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002F9TAKL8_CjhqSeu.webp",{"name":508,"bio":509},"flyotlin","誤打誤撞進入 crypto 世界，一個興趣使然的區塊鏈開發者。目前在 web3 fintech 公司打工中。",{"name":508,"bio":509},{"title":512,"describe":513,"type":30},"從近期駭客事件看 Cross-chain Bridge 的安全模型","Cross-Chain Bridge 是多鏈生態的重要基礎設施，但由近期多起案例都能發現，即使合約通過 audit，也不代表整體系統安全。\r\n此次分享將從近期案例（特別是 Kelp DAO 事件）出發，介紹目前常見的 Cross-Chain Bridge 設計模式，並且從系統安全角度拆解常見安全風險來源，帶大家深入認識 bridge security。",{"title":515,"describe":516,"type":30},"Understanding Cross-Chain Bridge Security Models Through Recent Hacks","Cross-chain bridges are critical infrastructure in the multi-chain ecosystem. However, recent incidents have shown that even if smart contracts pass audits, it does not guarantee the security of the overall system.\r\n\r\nIn this talk, we will start from recent cases (particularly the Kelp DAO incident) to introduce common cross-chain bridge design patterns. From a system security perspective, we will break down typical sources of risk and provide a deeper understanding of bridge security.",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002F9RGETL",{"id":520,"room":521,"start":12,"end":382,"language":70,"track":523,"speakers":527,"zh":535,"en":538,"tags":539,"uri":540},"9T8PEK",{"en":522,"zh-hant":10},"TR515",{"id":524,"name":525},527,{"en":526,"zh-hant":10},"Open Source Firmware",[528],{"id":529,"avatar":530,"zh":531,"en":534},"MLHFHG","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002Fface_246KSFN.jpeg",{"name":532,"bio":533},"Daniel Schaefer","System Software Engineer in Taiwan\r\n\r\nhttps:\u002F\u002Fgithub.com\u002FJohnAZoidberg",{"name":532,"bio":533},{"title":536,"describe":537,"type":30},"First coreboot port - Easier than expected","Demoing my first coreboot port and describing the experience. How to start, how to debug, common issues and how to contribute to upstream.",{"title":536,"describe":537,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002F9T8PEK",{"id":542,"room":543,"start":216,"end":239,"language":96,"track":545,"speakers":549,"zh":557,"en":560,"tags":563,"uri":564},"A3GRQD",{"en":544,"zh-hant":10},"TR313",{"id":546,"name":547},524,{"en":548,"zh-hant":548},"Hackers In Taiwan",[550],{"id":551,"avatar":552,"zh":553,"en":556},"UPUWGX","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FUPUWGX_FFx29Xk.webp",{"name":554,"bio":555},"David Su","---\r\n\r\nDavid Su 現職資安工程師，專注於 DevSecOps 架構設計、SSDLC 流程建置與資安合規實踐，致力於將資安融入企業的開發流程與組織文化。曾任資安顧問，累積滲透測試、GRC 審計與雲端資安等多元實務經驗。自 2018 年起持續深耕 HITCON 社群，參與活動組，長期投入資安社群活動規劃與技術交流推動。\r\n\r\n---\r\n\r\nDavid Su is a Security Engineer focused on DevSecOps architecture, SSDLC implementation, and security compliance — dedicated to embedding security into enterprise development workflows and organizational culture. He previously worked as a security consultant, building hands-on experience across penetration testing, GRC auditing, and cloud security. Since 2018, he has been an active member of the HITCON community, contributing to the Events team and driving technical exchange within the security community.",{"name":554,"bio":555},{"title":558,"describe":559,"type":30},"讓漏洞自己結案：純開源工具鏈打造 DevSecOps 閉環實戰","掃描工具裝了、CI\u002FCD 跑了，漏洞清單每天在長，但誰在追蹤？誰在修？修完了誰來確認？光有掃描工具不夠，挑戰在「掃完之後」：碎片化報告沒人聚合、供應鏈黑箱、沒有閉環。\r\n\r\n本場分享一套全程基於開源工具的 DevSecOps 閉環架構，涵蓋完整 SSDLC：以 OWASP Threat Dragon 進行威脅建模、OpenGrep 做 SAST、Trivy 產出 SBOM，搭配 DefectDojo 與 Dependency-Track 進行漏洞聚合與供應鏈監控，並自動開單至 Jira、GitHub 等平台，讓工程團隊在熟悉的介面處理資安問題。未來更可搭配 AI（如 Claude Code MCP）進行自動修補，讓閉環更進一步。\r\n\r\n除了介紹工具能做什麼，更會分享每個工具的真實限制，以及落地的三道高牆：開發團隊的抵制、SLA 永遠修不完的焦慮、跨部門推動阻力，以及如何用架構設計化解。\r\n\r\n漏洞從被掃出來，到自動開單、PR merge 後自動結案，RD 不需登入任何資安平台。SBOM 自動產出更讓 CRA 合規成為 B2B 競爭優勢。這是我們在實際專案中真實走過的一遍，所有工具均採用 OSI 認可授權。",{"title":561,"describe":562,"type":30},"Let Vulnerabilities Close Themselves: Building a DevSecOps Closed-Loop with a Pure Open Source Toolchain","Scanners run but vulnerabilities pile up — who tracks them, fixes them, and confirms they're closed?\r\n\r\nThis talk presents an open source DevSecOps closed-loop across the full SSDLC: Threat Dragon for threat modeling, OpenGrep for SAST, Trivy for SBOM generation, DefectDojo for aggregation and deduplication, and Dependency-Track for supply chain monitoring — all auto-ticketed to Jira\u002FGitHub so engineers never need to log into a security platform.\r\n\r\nWe go beyond tool introductions to share each tool's real limitations, and solutions to three adoption barriers: developer resistance, SLA overload, and cross-team pushback. SBOM automation also turns CRA compliance into a B2B competitive advantage.\r\n\r\nThis is a real-world project implementation. All tools are OSI-licensed open source.",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FA3GRQD",{"id":566,"room":567,"start":95,"end":40,"language":96,"track":568,"speakers":570,"zh":578,"en":581,"tags":582,"uri":583},"AKWMBV",{"en":544,"zh-hant":10},{"id":546,"name":569},{"en":548,"zh-hant":548},[571],{"id":572,"avatar":573,"zh":574,"en":577},"8BAALZ","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002F8BAALZ_BChhvlT.jpg",{"name":575,"bio":576},"scc","Architect @ Cycraft",{"name":575,"bio":576},{"title":579,"describe":580,"type":30},"Personal AI Assistant's Attack and Defense","This talk lays bare six months of intensive research, centered on OpenClaw. By benchmarking against industry counterparts like NVIDIA’s NeMo-Guardrails and Hermes-Agent, we provide insights that generalize across the entire personal AI assistant landscape.\r\n\r\nWe will conduct live demonstrations of five real-world attack vectors: Sandbox Backend Hijack, Subagent-to-Parent Injection, Group Chat History Pollution, Reply-chain Manipulation, and Supply Chain Attacks. Following the exploits, we’ll break down how to architect multi-layered defense-in-depth strategies. Finally, we will propose a unified \"SecurityResult\" interface: a standardized framework designed to let tools like NeMo Guardrails, LLM Guard, Presidio, Rebuff, and OpenGuardrails plug in seamlessly. Our goal is to move Personal AI Assistant security away from \"reinventing the wheel\" and toward a unified industry standard.",{"title":579,"describe":580,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FAKWMBV",{"id":585,"room":586,"start":239,"end":408,"language":96,"track":587,"speakers":589,"zh":595,"en":598,"tags":599,"uri":600},"ANYHFP",{"en":67,"zh-hant":10},{"id":72,"name":588},{"en":74,"zh-hant":74},[590],{"id":591,"avatar":3,"zh":592,"en":594},"XBZYFF",{"name":593,"bio":10},"Tim Hsu",{"name":593,"bio":10},{"title":596,"describe":597,"type":30},"200 行 Python 程式碼能訓練出什麼？ 詳解 Micro GPT","個 talk 會把我研究 Andrej Karpathy (open ai 創辦人） 的 micro-gpt 筆記走一遍。\r\nmicro-gpt 努力把Transformer、Attention、Embedding 等等元件用最簡單的方式實作，\r\n讓你用最少的負擔理解 gpt 的流程。\r\n\r\n或許...你之後就不會覺得 attention 這麼神秘了？ \r\n你也會知道漲價的記憶漲價被拿去裝什麼東西 :)",{"title":596,"describe":597,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FANYHFP",{"id":602,"room":603,"start":604,"end":605,"language":328,"track":606,"speakers":608,"zh":616,"en":619,"tags":620,"uri":621},"BBQDPR",{"en":39,"zh-hant":10},"2026-08-08T13:40:00+08:00","2026-08-08T14:10:00+08:00",{"id":384,"name":607},{"en":386,"zh-hant":10},[609],{"id":610,"avatar":611,"zh":612,"en":615},"8WM9UR","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002F8WM9UR_3t2PIg3.webp",{"name":613,"bio":614},"ryanycw(Ryan Wang)","An extra ordinary dev, who worked in DeFi, cares about privacy, tech, and ethereum (maybe). Won some ETHGlobal finalist & cohosting ETHTaipei. Lifestyle WIP: talk to me about triathlons, tennis, neo-jazz, food, and drinks.",{"name":613,"bio":614},{"title":617,"describe":618,"type":30},"The Privacy-preserving Identity Pipeline in KYC","KYC sits on a paradox: regulators demand strong identity proofs, but every byte of identity data a service stores becomes a permanent breach surface. The traditional answer: collect documents, run checks, file them away, turns every KYC provider into a high-value target and every user into a permanent disclosure.\r\n\r\nThis talk walks through an alternative pipeline now running in zkKYC field, where a stack of well-chosen cryptographic primitives composes into a system that proves claims about a user: their passport is valid, they are over 18, they are the same person who registered last month, without the server ever seeing their identity data. The journey spans device-bound keys, encrypted identity storage, government-signed document chains, biometric commitments, zero-knowledge proofs, and authenticated claim delivery but the goal throughout is to show how these pieces fit together, not to dwell on any single one.\r\n\r\nDesigned for engineers and product folks alike, the talk explains each primitive at a high level (no math), focuses on system design and uses architecture diagrams throughout. By the end, you will follow how the same passport scan that validates a user's age also produces a server-blind, unlinkable proof — and why this is starting to look like the only sustainable way to do identity at scale.\r\n\r\nTakeaways: a vocabulary for evaluating privacy-preserving identity systems, a concrete reference architecture, and an honest list of open problems (biometric revocation, post-quantum SNARKs, deduplication without linkability) this design has not yet solved.",{"title":617,"describe":618,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FBBQDPR",{"id":623,"room":624,"start":382,"end":453,"language":70,"track":625,"speakers":627,"zh":635,"en":638,"tags":639,"uri":640},"BKCVJK",{"en":237,"zh-hant":10},{"id":241,"name":626},{"en":243,"zh-hant":243},[628],{"id":629,"avatar":630,"zh":631,"en":634},"DBUY9H","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FDBUY9H_LrLFMzi.webp",{"name":632,"bio":633},"Nebiyu Elias Taelfe","Full-Stack Engineer with 7+ years of experience building, deploying, and maintaining web and mobile applications\r\nacross diverse domains. Skilled in Node.js, Express, Ruby on Rails, and PostgreSQL, with a track record of optimiz-\r\ning system performance and delivering scalable, production-ready solutions.",{"name":632,"bio":633},{"title":636,"describe":637,"type":30},"Building Native Mobile Apps Using Rails","We all know Rails is a very mature framework for building web applications? But can it be used to build native mobile apps as well? Remember what react native\u002Fflutter did to building mobile apps. You write your app once in those frameworks and magically your app works both in the App Store and Play Store. This significantly reduces the development effort and allow you to have a smaller team size. But what if I tell you can pretty much do the same thing using Rails. With this approach you don't maintain separate apps for your mobile and backend. You also take advantages of the power of Rails from the start. We will explore [hotwire native](https:\u002F\u002Fnative.hotwired.dev\u002F) and how it can be used to build mobile apps that are native. This way you can accelerate your mobile app development and cut down your development effort significantly. We will also explore how to use [strada](https:\u002F\u002Fstrada.hotwired.dev\u002F) to build native components using web technologies. I will also share my [starter template](https:\u002F\u002Fgithub.com\u002Fnebiyuelias1\u002Fturbo-rails-react-native-starter) for creating a mobile app project using Rails.",{"title":636,"describe":637,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FBKCVJK",{"id":642,"room":643,"start":604,"end":605,"language":328,"track":644,"speakers":646,"zh":654,"en":657,"tags":658,"uri":659},"BMENSU",{"en":326,"zh-hant":10},{"id":330,"name":645},{"en":332,"zh-hant":332},[647],{"id":648,"avatar":649,"zh":650,"en":653},"APK7CY","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FAPK7CY_qwfSraB.webp",{"name":651,"bio":652},"Madhur Jain","I am a software engineer at Canonical, working on the infrastructure and full-stack codebase behind ubuntu.com and canonical.com. I hold a B.Tech in Computer Science from IIT Bhilai (2024, CGPA 9.16, Institute Gold Medal) and have worked across high-stakes domains: retention infrastructure for 100 million users at PocketFM, a national-scale IAM system for India's 2024 General Elections, and open source contributions through Google Summer of Code at Submitty. I have spoken at Google DevFest Bhilai and Canonical-organised sprints.",{"name":651,"bio":652},{"title":655,"describe":656,"type":30},"Juju Does Everything. Here Is What That Is Like.","Juju is operator knowledge in code. It handles provisioning, K8s deployment, and automatic service wiring in one tool. This talk is about what that looks like in production at Canonical.\r\n\r\nMore info: https:\u002F\u002Fevents.canonical.com\u002Fevent\u002F146\u002Fcontributions\u002F940\u002F",{"title":655,"describe":656,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FBMENSU",{"id":661,"room":662,"start":663,"end":664,"language":96,"track":665,"speakers":667,"zh":675,"en":678,"tags":679,"uri":680},"BRNRVS",{"en":67,"zh-hant":10},"2026-08-08T16:10:00+08:00","2026-08-08T16:40:00+08:00",{"id":479,"name":666},{"en":481,"zh-hant":481},[668],{"id":669,"avatar":670,"zh":671,"en":674},"VLD79B","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FVLD79B_tGakcEq.webp",{"name":672,"bio":673},"Summer Lin","Hello, 我是Summer. 目前在Smartbox technology擔任firmware engineer, 韌體開發主要以zephyr RTOS為主. Smartbox technology 為有溝通障礙的殘障人士設計專用的電腦, 致力於深入研究使用者的需求, 開發出能幫助他們溝通的產品. 這對我來說是非常有意義的工作。\r\n在這之前我在台灣有10年的韌體開發經驗, 從最底層的韌體開發到Linux device driver 以及應用程式都有一些經驗, 希望我的演講可以帶給正在開發路上的你有點幫助.",{"name":672,"bio":673},{"title":676,"describe":677,"type":30},"Streaming Images from PC to an LCD on Zephyr over USB","Many embedded display systems rely on tightly coupled firmware or vendor-specific tooling, making it difficult to integrate with desktop applications or iterate quickly on UI workflows.\r\n\r\nIn this talk, I will present a practical end-to-end pipeline for streaming images from a host computer to an embedded LCD display, At its core is running open-source Zephyr RTOS. The system uses a custom USB protocol transfers, enabling efficient and flexible communication between a PC-side application (implemented in C#).\r\n\r\nI will walk through the system architecture, including USB protocol design, framebuffer handling, and integration with Zephyr’s display API, along with practical development experiences such as debugging, transfer performance, and reliability considerations, and approaches that can also be extended to lower-power display technologies such as e-paper systems.\r\n\r\nThe talk focuses on reusable design approaches and avoids proprietary implementation details. Attendees will gain a reusable model for building similar host-to-device display pipelines in their own projects.",{"title":676,"describe":677,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FBRNRVS",{"id":682,"room":683,"start":476,"end":477,"language":96,"track":685,"speakers":690,"zh":698,"en":701,"tags":704,"uri":705},"BWJMVT",{"en":684,"zh-hant":10},"TR209",{"id":686,"name":687},526,{"en":688,"zh-hant":689},"Open Source Policy","開源政策",[691],{"id":692,"avatar":693,"zh":694,"en":697},"M7GBZU","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FM7GBZU_tw6iEgo.webp",{"name":695,"bio":696},"RR","台中海線女孩，但總是被台北人當台南人，台南人當台北人。跨界公共關係、法律和服務設計，長期藏身於公務機關，近年進入博士班，日夜焚香祝禱能夠順利畢業。身為連程式碼都沒看過的麻瓜，2015 年誤打誤撞進入開源和 g0v 社群，歷經 cofacts 和 g0v Summit 一路停留至今，最後來到 OCF。",{"name":695,"bio":696},{"title":699,"describe":700,"type":30},"從草莽到廟堂：開放原始碼如何走進歐盟和聯合國","2000 年時，FOSDEM只是一個小型純技術聚。但伴隨著歐盟對於「數位韌性（Digital Resilience）」的重視，各種圍繞 FOSDEM 的開放原始碼政策辯論活動開始出現，並在 2019 年出現了首次由歐盟官方舉辦的政策會議 EU Open Source Policy Meeting。2020 年起，升級為 EU Open Source Policy Summit 更帶頭串連出了 EU Open Source Week。至此每年 2 月的布魯塞爾不再僅有技術宅宅，更聚集了歐洲的資訊官員與廠商。\r\n\r\n2025年，EU Openn Source Week 的成功案例吸引了聯合國的注意，更深信開放原始碼透明、安全、去中心化的精神正是解決全球數位落差、實現永續發展目標（Sustainable Development Goals，SDGs） 的理想工具，UN Open Source Week 也應運而生。\r\n\r\n到底這一切具體是怎麼發生的？這一切對於社群和各國政府及廠商將帶來怎麼樣的轉變，我們讓今年特別前往第一線參與的雁子、 RR 為您帶來第一手資訊！",{"title":702,"describe":703,"type":30},"From Streets to Stars: How Open Source Conquered the EU and the UN","FOSDEM started with a group of engineers sharing open technologies. However, as policy debates toward digital resilience started cropping up around the fringes of FOSDEM, these side events evolved into the EU Open Source Policy Summit, sparking the creation of EU Open Source Week. Now, every February, Brussels isn't just full of hackers in hoodies—it’s a hub for Europe’s top CIOs, policymakers, and industry leaders.\r\n\r\nFast forward to 2025: the success of EU Open Source Week caught the attention of the United Nations. The UN recognized that the tenets of Open Source—transparency, security, and decentralization—enables it to bridge the global digital divide and achieve Sustainable Development Goals (SDGs). Thus, UN Open Source Week was born.\r\n\r\nHow did open source get here? What does this institutional shift mean for the community, governments, and the industry at large? Join Yanzi and RR, who were on the ground this year, as they bring you the latest first-hand information!",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FBWJMVT",{"id":707,"room":708,"start":123,"end":709,"language":96,"track":710,"speakers":712,"zh":720,"en":723,"tags":726,"uri":727},"BXTFW9",{"en":121,"zh-hant":10},"2026-08-08T10:50:00+08:00",{"id":125,"name":711},{"en":127,"zh-hant":128},[713],{"id":714,"avatar":715,"zh":716,"en":719},"WYBP8C","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FWYBP8C_UyPUjjG.webp",{"name":717,"bio":718},"Cui Jia Wei","崔家瑋目前就讀於台灣大學法律研究所，研究領域為開源人工智慧治理。他於 2022 年通過律師與司法官考試，在台灣最大的虛擬資產交易所擔任法律專員。在學術與工作之外，他亦參與台灣的非政府組織與公民科技社群。他所參與的公民科技專案 vTaiwan 探討數位工具應用於審議民主的可能性。vTaiwan 在 2023 年與 OpenAI 合作，探討利用審議民主改善人工智慧治理的提案；自 2024 年開始，vTaiwan 與台灣網路資訊中心合作，以數位工具協助網路與人工智慧治理的多方利害關係人討論。他於 2025 年亦獲選成為 TWNIC Academy Fellow。\r\nCui Jia-Wei is currently a graduate student at the Graduate Institute of Law at National Taiwan University, where his research focuses on the governance of open-source artificial intelligence. He passed both the bar examination and the judicial officer examination in 2022, and currently works as a legal specialist at Taiwan’s largest virtual asset exchange.\r\n\r\nIn addition to his academic and professional work, he is actively involved in non-governmental organizations and civic tech communities in Taiwan. He participates in the civic tech project vTaiwan, which explores the use of digital tools in deliberative democracy. In 2023, vTaiwan collaborated with OpenAI to develop proposals on improving AI governance through deliberative democracy. Since 2024, vTaiwan has partnered with the Taiwan Network Information Center to facilitate multi-stakeholder discussions on internet and AI governance using digital tools. In 2025, he was selected as a TWNIC Academy Fellow.",{"name":717,"bio":718},{"title":721,"describe":722,"type":30},"開源人工智慧的定義、風險與治理","從 DeepSeek 在 2025 年春節橫空出世以來，有關開源人工智慧的討論正式從技術與開源社群進入公眾視野。然而開源社群與人工智慧的發展其實比 DeepSeek 引發的討論更加深遠。有許多現在人工智慧技術仰賴的訓練架構以開源釋出，甚至開源社群與自由軟體的起源也與 MIT 的人工智慧實驗室相關。近期，開源人工智慧又因為主權 AI 的討論，而成為不同國家關注的焦點。在這樣的背景下，本議程將會聚焦在開源人工智慧的定義、風險討論與各國政策比較上。\r\n\r\n首先，開源人工智慧所涵蓋的人工智慧技術究竟為何？只要公開模型程式碼，就可以稱作開源人工智慧了嗎？人工智慧的資料權重，甚至訓練的資料集是否要一併公開，才能算是開源人工智慧？各家大廠所宣稱的開源人工智慧真的是開源嗎？\r\n其次，開源人工智慧由於其容易散布的特性，相較於閉源的人工智慧是否會產生新的風險？抑或是開源所帶來的透明度反而可以作為風險管理的手段與工具？\r\n再者，各國對於開源人工智慧的態度為何？歐盟在《人工智慧法》中，明確規定以免費與開源方式釋出的人工智慧組件，可以不受該法規範。然而該豁免是否有例外與限制？美國政府對於開源人工智慧的態度為何，特別是在中美競爭逐漸蔓延到科技上的競逐，而中國在多款開源人工智慧模型的釋出均佔有主導地位時，美國會如何應對？新加坡對於開源人工智慧的政策又是如何？\r\n以及最後，台灣可以從這些討論中，獲得什麼啟發呢？",{"title":724,"describe":725,"type":30},"Open-Source AI: Definition, Risk and Governance","Since DeepSeek burst into the spotlight during the 2025 Lunar New Year, discussions around open-source artificial intelligence have moved from technical and open-source communities into the public eye. However, the relationship between open-source communities and the development of AI runs far deeper than the recent attention triggered by DeepSeek. Many of the training frameworks that underpin today’s AI technologies have been released as open source, and the origins of both the open-source and free software movements are closely tied to the MIT AI Lab. More recently, open-source AI has also become a focal point for national attention due to the rise of “sovereign AI” debates. Against this backdrop, this session will focus on the definition of open-source AI, its associated risks, and a comparative analysis of national policies.\r\nFirst, what exactly counts as “open-source AI”? Is releasing model code alone sufficient to qualify as open source? Should model weights, or even the training datasets, also be made publicly available? And are the so-called “open-source” AI systems promoted by major tech companies truly open?\r\nSecond, given the ease with which open-source AI can be disseminated, does it introduce new risks compared to closed-source systems? Or, alternatively, can the transparency enabled by open source serve as a tool for risk management and mitigation?\r\nThird, how are different countries approaching open-source AI? The European Union, through the EU Artificial Intelligence Act, explicitly provides that AI components released in a free and open-source manner may be exempt from certain regulatory requirements. However, are there exceptions or limitations to this exemption? What is the stance of the United States government, particularly as technological competition between the U.S. and China intensifies, and China takes a leading role in releasing multiple open-source AI models? How is Singapore positioning its policies toward open-source AI?\r\nFinally, what lessons can Taiwan draw from these global discussions?",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FBXTFW9",{"id":729,"room":730,"start":454,"end":731,"language":96,"track":732,"speakers":736,"zh":740,"en":743,"tags":744,"uri":745},"CHQFP8",{"en":121,"zh-hant":10},"2026-08-08T15:20:00+08:00",{"id":733,"name":734},525,{"en":735,"zh-hant":735},"AI x Civic Tech",[737],{"id":145,"avatar":146,"zh":738,"en":739},{"name":148,"bio":149},{"name":148,"bio":149},{"title":741,"describe":742,"type":30},"生成式 AI 時代的公民開發治理：從業務創新到企業納管","當 AI 讓業務端也能快速做出系統時，企業或公民科技社群要怎麼定義邊界、責任與收編機制。\r\nAI 公民開發該鼓勵創新，還是優先納管？\r\n收編還是放生？企業如何面對 AI 自發開發潮\r\nAI 生成系統進企業後，資訊部門是加速器還是防火牆？\r\n公民科技社群的氛圍產物像是美麗煙火，天亮後我們的未來。",{"title":741,"describe":742,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FCHQFP8",{"id":747,"room":748,"start":453,"end":94,"language":96,"track":749,"speakers":751,"zh":759,"en":762,"tags":764,"uri":765},"CUHVKJ",{"en":121,"zh-hant":10},{"id":733,"name":750},{"en":735,"zh-hant":735},[752],{"id":753,"avatar":754,"zh":755,"en":758},"JQN3VV","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FJQN3VV_jSJUX8N.webp",{"name":756,"bio":757},"蕭新晟","蕭新晟為應援科技（Oen Tech）創辦人、執行長暨董事長，長期投入金融科技、支付基礎設施與數位服務創新。應援科技專注於整合第三方支付、CRM、電子發票、訂閱與票務等能力，服務 NPO、創作者、宗教、政治與各類組織，協助建立更完整的數位支持與金流生態。蕭新晟具美國新創與工程背景，亦長期參與公民科技與開放政府倡議，關注 AI、金流科技與制度創新的交會，近年特別聚焦 AI 在支付體驗、營運效率與金融服務流程中的實際應用。",{"name":756,"bio":757},{"title":760,"describe":761,"type":30},"AI Sovereignty in Practice: From Taiwan Benchmark to Production Payment Agents","如果 AI agent 可以查支付資料、追金流、判讀客服案件，我們要如何知道它沒有帶著不可見的模型過濾、資料權限漏洞或不可稽核 workflow 在做決定？\r\n\r\n 本講分享 OEN 將 agents 導入 production payment infrastructure 的工程經驗：先用 Taiwan Sovereignty Benchmark 審計中文模型在台灣主權問題上的回答、過濾與 system prompt bypass；再用 payment-mcp + payment-skills 把 payment data access 與 charge lookup、phone trail、Twilio\u002FSMS、payout failure diagnosis 等流程變成 scoped、logged、可 review、可 fork 的 artifacts。最後連回台灣 AI 資料利用法制：支持善意、可稽核、非替代市場的創新，但避免授權制度成為新資料壟斷。",{"title":760,"describe":763,"type":30},"How can we audit an AI agent before letting it inspect payment data, trace transactions, or help diagnose customer-support cases? This talk shares OEN's engineering path from Taiwan Sovereignty Benchmark to production payment agents: first auditing Chinese-language models for Taiwan-sovereignty filtering and system-prompt bypass behavior, then using payment-mcp and payment-skills to make payment-data access and workflows scoped, logged, reviewable, and forkable. It closes by connecting this engineering boundary to Taiwan's AI data-use policy: legal certainty should protect good-faith, auditable, non-market-substituting innovation without turning authorization into a new data monopoly.",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FCUHVKJ",{"id":767,"room":768,"start":500,"end":11,"language":96,"track":769,"speakers":771,"zh":779,"en":782,"tags":785,"uri":786},"CUVVTA",{"en":67,"zh-hant":10},{"id":72,"name":770},{"en":74,"zh-hant":74},[772],{"id":773,"avatar":774,"zh":775,"en":778},"W377FQ","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FW377FQ_xpXKZTz.webp",{"name":776,"bio":777},"小松 XiaoSong","我叫小松，本職擔任Devops 工程師，是2026 Devops Day Taipei 講者之一，曾是Google Developer Group on Campus Ex-Organizer，也是Google Devfest Taipei 2024 的Google 官方工作坊 講師之一",{"name":776,"bio":777},{"title":780,"describe":781,"type":30},"當 Release 不再靠人工：用 Python 生態打造自動化版本治理與 DevX 實戰","在許多 Python 專案中，Release 流程仍高度依賴人工：開發者需要手動撰寫 Changelog、整理版本差異，前後端或不同服務之間的版本資訊也常缺乏一致性，導致 PM、QA、RD 之間溝通成本高，並增加部署與回溯的困難度。\r\n\r\n本次分享將以 Python 專案開發流程 為背景，介紹如何結合 Conventional Commits、分支治理策略（GitFlow \u002F Trunk-based）與 CI\u002FCD 自動化流程，打造一套「不依賴人工」的版本治理機制。\r\n\r\n我們會展示如何在 Python 專案中：\r\n\r\n透過 commit 規範自動驅動版本號（符合 semantic versioning）\r\n使用工具（如 semantic-release、commitizen、Python packaging 生態）自動產生 Release Note\r\n將版本資訊整合進部署流程（如 artifact \u002F container image）\r\n建立可追蹤、可回溯、可快速切換的版本管理方式\r\n\r\n本分享將以實際 Python 專案與 CI\u002FCD pipeline 為例，展示從 commit 到 release 的完整自動化流程。\r\n\r\n本分享強調以 Python 生態為核心，但方法可延伸至其他技術棧。透過這套實戰流程，Release 不再是手動整理的負擔，而是可持續、自動化的工程能力，進一步提升 Developer Experience（DevX）與團隊協作效率。",{"title":783,"describe":784,"type":30},"No More Manual Releases: Automating Versioning and DevX in Python Projects","In many Python projects, the release process still relies heavily on manual work—writing changelogs, tracking version differences, and coordinating across teams. This often leads to inconsistent versioning and high communication costs between developers, QA, and product managers.\r\n\r\nThis talk demonstrates how to build an automated release workflow in Python projects using Conventional Commits, CI\u002FCD pipelines, and open-source tools such as semantic-release and commitizen. We will show how to generate release notes automatically, align versioning across services, and create a traceable and reproducible release process.\r\n\r\nBy turning release into an automated system rather than a manual task, teams can significantly improve developer experience (DevX), reduce friction, and enhance collaboration.",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FCUVVTA",{"id":788,"room":789,"start":605,"end":476,"language":96,"track":790,"speakers":792,"zh":800,"en":803,"tags":806,"uri":807},"CV7GX9",{"en":407,"zh-hant":10},{"id":410,"name":791},{"en":412,"zh-hant":412},[793],{"id":794,"avatar":795,"zh":796,"en":799},"QPZCZG","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FQPZCZG_ve97ewF.webp",{"name":797,"bio":798},"許智庭 Tim Hsu","現為 AI 策略規劃師與陪跑教練，協助客戶盤點需求痛點以及協助 AI 技術選型，藉由實際了解客戶需求提供適合的 AI 解決方案，並且具備五年以上前端開發經驗與七年平面設計背景，能兼具使用者體驗與程式實作。在職涯中曾參與銀行、保險、航空與電商等不同產業的系統開發，熟悉大型專案的資料流程、API 串接、跨瀏覽器調整與前後端協作模式。 具備一般商務應用與程式開發雙棲等實務經歷，讓我能站在 AI 這個巨人的肩膀上看得更遠。",{"name":797,"bio":798},{"title":801,"describe":802,"type":30},"從政府 Open Data 到城市探索網站：用 Nuxt 3、Codex 與 GitHub Actions 打造北北桃中公園地圖","政府開放資料雖然公開，但往往分散在不同縣市、欄位格式不一致、資料完整度也不同，距離「一般使用者真的能用」還有一段距離。\r\n\r\n本場分享將以一個實際開源專案為例，說明如何使用 Nuxt 3、Vue 3、TypeScript 將臺北市、新北市、桃園市、臺中市的公園 Open Data 整合成統一資料模型，並實作搜尋、篩選、排序、地圖標示與防災資訊查詢等功能。\r\n\r\n分享中也會介紹如何透過 Codex 作為 pair programmer，協助理解資料欄位、建立正規化流程、拆分前端元件與加速開發迭代；以及如何使用 GitHub Actions 自動更新部分資料來源並部署到 GitHub Pages。\r\n\r\nOpen Data 的價值不只在於公開，更在於如何透過 JavaScript 生態系與開源工具，將資料轉化成可探索、可維護、可延伸的實用產品。",{"title":804,"describe":805,"type":30},"From Government Open Data to an Urban Exploration Website: Building a Park Map for Taipei, New Taipei, Taoyuan, and Taichung with Nuxt 3, Codex, and GitHub Actions","Government open data is publicly available, but it is often scattered across different cities and counties, with inconsistent field formats and varying levels of data completeness. There is still a gap between “data being open” and “data being truly usable by everyday users.”\r\n\r\nIn this talk, I will use a real open source project as an example to demonstrate how Nuxt 3, Vue 3, and TypeScript can be used to integrate park open data from Taipei City, New Taipei City, Taoyuan City, and Taichung City into a unified data model. The project implements features such as search, filtering, sorting, map visualization, and disaster-prevention park information lookup.\r\n\r\nI will also share how Codex was used as a pair programmer to help understand data fields, build data normalization workflows, split frontend components, and accelerate development iterations. In addition, I will explain how GitHub Actions can be used to automatically update selected data sources and deploy the site to GitHub Pages.\r\n\r\nThe value of Open Data is not only in making data public, but also in using the JavaScript ecosystem and open source tools to transform that data into practical products that are explorable, maintainable, and extensible.",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FCV7GX9",{"id":809,"room":810,"start":477,"end":40,"language":70,"track":811,"speakers":813,"zh":821,"en":824,"tags":825,"uri":826},"CVHJQU",{"en":67,"zh-hant":10},{"id":479,"name":812},{"en":481,"zh-hant":481},[814],{"id":815,"avatar":816,"zh":817,"en":820},"URL8AA","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FURL8AA_yjhFhy7.webp",{"name":818,"bio":819},"Leon Anavi","Leon Anavi is an open source enthusiast and a senior software engineer at Konsulko Group. He is an active contributor to various Yocto\u002FOpenEmbedded meta layers, Automotive Grade Linux (AGL) and many other open source projects. His professional experience includes web and mobile application development for various platforms as well as porting and maintaining embedded Linux distributions to Raspberry Pi and devices with x86-64, i.MX6, i.MX8, NVIDIA Tegra, RISC-V, Amlogic, Rockchip and Allwinner (aka sunxi) SoC. Leon holds a masters in Information Technology from the Technical University Sofia. His previous speaking experience includes talks about open source software and hardware during virtual and in-person events in San Francisco, San Diego, Portland (OR), Hong Kong, Shanghai, Shenzhen, Brussels, Lyon, Berlin, Edinburgh, London, Cambridge, Bratislava, Prague, Nice, Sofia and his hometown Plovdiv.",{"name":818,"bio":819},{"title":822,"describe":823,"type":30},"Smart Home Displays with ePaper: A Maker’s Guide Using ESP32 and the Arduino Ecosystem","ePaper displays are an ideal solution for always-on low power smart home interfaces, offering excellent readability and near zero energy consumption when static. In this session, we show how makers can build their own smart home displays using ESP32 based XIAO modules and the Arduino ecosystem.\r\n\r\nStarting from the basics, you will learn how to interface a 4.2\" monochrome ePaper display driven by the SSD1680 controller using the open source Seeed GFX Library. The presentation covers hardware comparison of the XIAO ePaper Display breakout boards and explains their differences and typical use cases. The session demonstrates how to render dynamic content such as sensor data, home automation status information, and custom dashboards while optimizing for low power operation. By the end, you will have a practical foundation for building your own battery powered smart home displays suitable for digital signage, room dashboards, and portable information panels.\r\n\r\nThis talk is suitable for everyone and no previous experience is required. It will be hopefully useful both for engineers and makers who are interested in using ePaper displays in their projects, whether they are just getting started or looking to expand their existing skills.",{"title":822,"describe":823,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FCVHJQU",{"id":828,"room":829,"start":731,"end":830,"language":96,"track":831,"speakers":833,"zh":841,"en":844,"tags":847,"uri":848},"DAPVAM",{"en":121,"zh-hant":10},"2026-08-08T15:50:00+08:00",{"id":733,"name":832},{"en":735,"zh-hant":735},[834],{"id":835,"avatar":836,"zh":837,"en":840},"93JF3H","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002F93JF3H_970IERa.webp",{"name":838,"bio":839},"Josh (杜佳憲)","任職於消防前線單位，長期關注科技救災、公共服務創新與公民科技（Civic Tech）領域。致力於將現代生成式 AI 技術、自動化流程與開源精神導入傳統公務體系。曾實質參與數位發展部公民科技試驗場域計畫，期盼透過跨領域的公私協力，為基層消防單位打造具備數位韌性的智慧化系統，讓科技真正赋能公共服務。\r\n\r\nA firefighter dedicated to disaster response technology and Civic Tech. Holding a Master's degree in Environmental Economic Information Management, he focuses on integrating generative AI, workflow automation, and open-source principles into the public sector to build digitally resilient systems through public-private collaboration.",{"name":838,"bio":839},{"title":842,"describe":843,"type":30},"讓人力回歸前線：智慧排程系統的 AI 輔助開發與開源公私協力實錄","【目標聽眾】\r\n本演講適合關心公民科技（Civic Tech）、政府科技（GovTech）、流程自動化，以及對公私協力推動體制數位轉型感興趣的開發者與推廣者。\r\n\r\n【痛點與背景】\r\n公部門（特別是偏鄉單位）常面臨人力吃緊的挑戰。臺東縣消防局每年需處理密集的消防安全設備紙本申報、稽核與複查排程，繁瑣且重複的行政負擔極大地排擠了前線搜救與訓練的精力。\r\n\r\n【解法與開源實踐】\r\n為了解決此痛點，臺東縣消防局積極導入數位發展部「公民科技試驗場域」計畫，與公民科技團隊度合作，成功開發出「申報自動排程系統」。本專案將原本繁瑣的公務行政流轉化為智慧化自動排程演算法，落實公私協力與開源共好精神。\r\n\r\n【議程大綱 (30分鐘)】\r\n- [00-05m] 偏鄉消防的數位痛點：當有限人力遇上每年萬件申報案\r\n- [05-15m] 流程智慧化：自動排程系統的架構設計、法規邏輯轉化與開源協作\r\n- [15-25m] 落地成果與踩坑經驗：Civic Tech 團隊與公部門跨域溝通的淚與笑\r\n- [25-30m] Q&A 與未來展望：如何複製此模式至更多公共服務場域",{"title":845,"describe":846,"type":30},"Returning Rescuers to the Frontline: AI-Assisted Development and Open Public-Private Collaboration for Fire Safety Equipment Smart Scheduling System","[Target Audience]\r\nDevelopers, public sector tech enthusiasts, and open-source advocates interested in Civic Tech, Gen-AI assisted development, and workflow automation.\r\n\r\n[Pain Points & Context]\r\nPublic sectors often face severe labor shortages. The Taitung County Fire Bureau handles nearly 10,000 fire safety equipment declarations and inspection schedules annually. This tedious administrative burden heavily distracts personnel from frontline rescue and training.\r\n\r\n[AI Application & Open Source Practice]\r\nThrough the \"Civic Tech Sandbox\" initiative by the Ministry of Digital Affairs, the Fire Bureau collaborated with civic tech team \"CivicFlow\" to build a \"Smart Scheduling System.\" A key highlight is the heavy utilization of Gen-AI tools (such as Cursor and LLMs) during development to translate complex regulations into automated algorithms, showcasing the potential of Gen-AI in empowering Civic Tech and fostering public-private open collaboration.\r\n\r\n[Outline (30 mins)]\r\n- [00-05m] Pain points of rural firefighting administrative overload.\r\n- [05-15m] AI-powered Civic Tech: Translating regulations into open algorithms via LLMs.\r\n- [15-25m] Implementation & Lessons: Public-private collaboration in a government sandbox.\r\n- [25-30m] Q&A and Future: What's next for Gen-AI in public services.",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FDAPVAM",{"id":850,"room":851,"start":454,"end":731,"language":96,"track":852,"speakers":854,"zh":869,"en":872,"tags":875,"uri":876},"DYJUUF",{"en":192,"zh-hant":10},{"id":196,"name":853},{"en":198,"zh-hant":198},[855,862],{"id":856,"avatar":857,"zh":858,"en":861},"MGQQE7","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FMGQQE7_7SPwGDu.webp",{"name":859,"bio":860},"William Hong","One of CCU OSLAB team member\r\n\r\nA newbie of Linux kernel, and hope my contributions to Linux kernel can make this world better",{"name":859,"bio":860},{"id":863,"avatar":864,"zh":865,"en":868},"M3CM9H","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FM3CM9H_jEHVgY6.jpg",{"name":866,"bio":867},"曾俊瑋","CCU OSLAB",{"name":866,"bio":867},{"title":870,"describe":871,"type":30},"從 Test-and-Set 到 NUMA-aware Locks：重新思考現代硬體上的 Spinlocks","現代多核心系統高度仰賴高效率的同步機制，但許多開發者仍將 spinlock 視為黑盒子。\r\n在本場演講中，我們將探討 spinlock 的設計如何演進，以因應現代硬體上的可擴展性挑戰。從經典的 test-and-set 與 ticket locks 出發，我們會檢視它們在高競爭情境下的限制，並說明以queue為基礎的方法，例如 Linux kernel 的 qspinlock，如何確保公平性並提升效能。\r\n接著，我們將進一步探討更進階的設計，例如 Compact NUMA-aware Locks（CNA），並介紹我們自己的實驗性實作 RON，強調快取一致性與 NUMA 效應所帶來的挑戰。\r\n我們也會分享 CNA 設計者所遭遇的挑戰，以及我們在嘗試將新的鎖設計 upstream 到 Linux kernel 過程中的經驗與教訓。\r\n最後，我們將討論在真實系統中提升效能的策略；相較於設計鎖，開發者更常面對的是如何使用鎖。這些策略包括降低鎖的大小，以及考慮 RCU 等替代同步機制。\r\n與會者將能獲得關於鎖可擴展性、效能瓶頸，以及如何分析現代多核心架構上同步行為。\r\n本場次延續我們先前的演講，但會更著重於 CNA 與 RON 的設計與評估，並納入 kernel maintainers 的觀點。我們將聚焦於真實世界中的取捨與決策，提供與會者可付諸行動的觀點，協助處理因鎖競爭所造成的效能瓶頸，並評估替代同步策略。",{"title":873,"describe":874,"type":30},"From Test-and-Set to NUMA-aware Locks: Rethinking Spinlocks on Modern Hardware","Modern multicore systems rely heavily on efficient synchronization, yet many developers treat spinlocks as a black box.\r\n\r\nIn this talk, we explore how spinlock designs evolve to address scalability challenges on modern hardware.\r\nStarting from classic test-and-set and ticket locks, we examine their limitations under contention and how queue-based approaches such as the Linux kernel qspinlock ensure fairness and improve performance.\r\n\r\nWe further examine advanced designs such as Compact NUMA-aware Locks (CNA) and present our own experimental implementation, RON,\r\nhighlighting the challenges of cache coherence and NUMA effects.\r\n\r\nWe also highlight the challenges encountered by the designers of CNA,\r\nas well as our experiences and lessons learned from attempting to upstream new lock designs into the Linux kernel.\r\n\r\nFinally, we discuss practical strategies for improving performance in real-world systems,\r\nwhere developers more often use locks than design them.\r\nThese include applying fine-grained locking techniques (i.e., reducing lock granularity) and considering alternative synchronization mechanisms like RCU.\r\n\r\nAttendees will gain practical insights into lock scalability, performance bottlenecks,\r\nand how to analyze synchronization behavior on modern multicore architectures. \r\n\r\nBuilding on our previous talk, this session places greater emphasis on the design and evaluation of CNA and RON, incorporating insights from kernel maintainers.\r\nWe focus on real-world trade-offs and practical decision-making in synchronization,\r\nproviding attendees with actionable perspectives on addressing performance bottlenecks caused by lock contention and considering alternative synchronization strategies.",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FDYJUUF",{"id":878,"room":879,"start":604,"end":605,"language":96,"track":880,"speakers":882,"zh":890,"en":893,"tags":894,"uri":895},"EHPZSF",{"en":522,"zh-hant":10},{"id":524,"name":881},{"en":526,"zh-hant":10},[883],{"id":884,"avatar":885,"zh":886,"en":889},"ENGZXD","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FENGZXD_o1vPZuC.jpg",{"name":887,"bio":888},"Ann Cheng","在Industry Server打雜多年的老妹。",{"name":887,"bio":888},{"title":891,"describe":892,"type":30},"Evolving Arm Firmware Ecosystem - Firmware Testing and Telemetry","Firmware today is not just about “it boots.” We also care about whether it’s correct, testable, and even observable at runtime.\r\nIn this talk, I’ll share some of Arm’s recent work in opensource, focusing on two areas: testing and telemetry.\r\n\r\n1. Firmware testing\r\n1.1 edk2-test (UEFI-SCT) and FWTS\r\n\r\n2. Firmware telemetry\r\n2.1 OpenBMC bulk telemetry",{"title":891,"describe":892,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FEHPZSF",{"id":897,"room":898,"start":238,"end":239,"language":70,"track":900,"speakers":905,"zh":913,"en":916,"tags":917,"uri":918},"ETFVF8",{"en":899,"zh-hant":10},"TR211",{"id":901,"name":902},509,{"en":903,"zh-hant":904},"Miscellaneous Open Source Topics","綜合議程 - 各種開源議題",[906],{"id":907,"avatar":908,"zh":909,"en":912},"LGJ8EX","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FLGJ8EX_jVD0ucz.webp",{"name":910,"bio":911},"Jim Grisanzio","Jim Grisanzio has built FOSS communities at Oracle and Sun Microsystems for 25 years. He’s interviewed hundreds of developers on Java, AI, Linux, Oracle Database, MySQL, Machine Learning, and Open Source for two separate podcasts: The Oracle Groundbreakers Podcast and also The Duke's Corner Java Podcast. He published two newsletters for Java and Oracle Database developers, produced developer videos and photography, facilitated Java User Group and Java Champions programs, and delivered his own community sessions and keynotes at conferences. He's created and run user groups, and managed developer interactions via social media platforms on YouTube, Facebook, LinkedIn, and Twitter\u002FX. Jim also served as the OpenSolaris Community Manager at Sun, and led communications for Sun’s Open Source projects and senior executives. Jim's currently a consultant and planning his new podcast called The Long Build. You can follow Jim on his website https:\u002F\u002Fgrisanzio.com\u002F.",{"name":910,"bio":911},{"title":914,"describe":915,"type":30},"Developers: Start Contributing Now!","The FOSS community thrives because it's built on collaboration, mentorship, and contribution. Drawing from hundreds of podcast interviews with FOSS developers globally, this session highlights how strategic contributions can create benefits for individual developers, corporate employers, and the entire community. Attendees will gain actionable strategies to start contributing effectively while learning how such efforts open doors to leadership, visibility, and long-term relationships within the community.",{"title":914,"describe":915,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FETFVF8",{"id":920,"room":921,"start":382,"end":453,"language":328,"track":923,"speakers":928,"zh":936,"en":939,"tags":940,"uri":941},"FCCAZW",{"en":922,"zh-hant":10},"TR210",{"id":924,"name":925},529,{"en":926,"zh-hant":927},"Open Source Business Ecosystem","開源商業模式",[929],{"id":930,"avatar":931,"zh":932,"en":935},"8KG7ST","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002F8KG7ST_lBZJAd6.webp",{"name":933,"bio":934},"Bruce Momjian","Bruce Momjian is co-founder and core team member of the PostgreSQL Global Development Group, and has worked on PostgreSQL since 1996. He has been employed by EDB since 2006. He has spoken at many international open-source conferences and is the author of PostgreSQL: Introduction and Concepts, published by Addison-Wesley. Prior to his involvement with PostgreSQL, Bruce worked as a consultant, developing custom database applications for some of the world's largest law firms. As an academic, Bruce holds a Masters in Education, an honorary doctorate, was a high school computer science teacher, and lectures internationally.",{"name":933,"bio":934},{"title":937,"describe":938,"type":30},"Will Postgres Live Forever?","This presentation explains how open source software can live for a very long time, and covers the differences between proprietary and open source software life cycles. It also covers the increased adoption of open source, and many of the ways that Postgres is innovating to continue to be relevant.",{"title":937,"describe":938,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FFCCAZW",{"id":943,"room":944,"start":709,"end":499,"language":328,"track":945,"speakers":950,"zh":958,"en":961,"tags":962,"uri":963},"FFEAJR",{"en":922,"zh-hant":10},{"id":946,"name":947},518,{"en":948,"zh-hant":949},"Taiwan MySQL User Group","台灣MySQL使用者社群",[951],{"id":952,"avatar":953,"zh":954,"en":957},"MPJLZF","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FMPJLZF_5lX3rA5.webp",{"name":955,"bio":956},"Ronald Bradford","Ronald Bradford is a well recognized open-source database community contributor member and MySQL subject matter expert. He has been presenting at database, open source and cloud computing events worldwide for 20 years. As a published author of four books in MySQL, Ronald has peer recognitions include MySQL Community Member of the Year, Oracle ACE Director, MySQL RockStar Legend and multiple certifications.\r\n\r\nHis enterprise data architecture experience for large-scale data systems, specializes in designing automated, resilient, and scalable data solutions. His hands-on approach transforms complex challenges into measurable business success for customers.",{"name":955,"bio":956},{"title":959,"describe":960,"type":30},"Planning your upgrade to MySQL 9.7","The recent release of MySQL  9.7 Long Term Support (LTS) community editions, joins MySQL 8.4 LTS as the two current LTS releases.  The popular and widely used MySQL 8.0 has now reached EOL. Organizations running MySQL 8.0 should be migrating to MySQL 8.4 and evaluating the roadmap of features and capabilities of the new MySQL 9.7.\r\n\r\nIn this session, we'll take a deep dive into the new capabilities shipping in MySQL 9.7 Community Edition: full DML support for JSON Duality Views, the Hypergraph Optimizer, JavaScript stored procedures, enhanced security management, and a number of Enterprise-only components now available to all — including replication observability tools and OpenTelemetry-based telemetry integration. \r\nThis talk focuses on how to responsibly evaluate and test a major version release before production deployment. We'll walk through a practical checklist covering compatibility assessment, workload testing, staged rollout strategies, and rollback planning to help developers integrating new features into application workflows, DBAs managing the migration and monitoring behavior changes, and engineering leadership weighing the timing and scope of the upgrade.\r\nAttendees will leave with a better understanding of MySQL 9.7's most impactful features and a strategy they can take directly back to their team to evaluate them.",{"title":959,"describe":960,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FFFEAJR",{"id":965,"room":966,"start":239,"end":408,"language":328,"track":967,"speakers":969,"zh":977,"en":980,"tags":981,"uri":982},"FKFEQJ",{"en":192,"zh-hant":10},{"id":196,"name":968},{"en":198,"zh-hant":198},[970],{"id":971,"avatar":972,"zh":973,"en":976},"QB3ZKV","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FQB3ZKV_MrX5dba.webp",{"name":974,"bio":975},"Hoyeon Lee","Hoyeon Lee is a BPF System Engineer at SUSE, focusing on BPF and system internals in the Linux kernel. He enjoys sharing knowledge through writing and speaking, and has been active in open source communities and technical translation.",{"name":974,"bio":975},{"title":978,"describe":979,"type":30},"Linux Kernel Tracing Internals","This session dives into the core mechanisms of Linux kernel tracing. \r\nWe explore dynamic instrumentation such as exception-based kprobe and code-patching ftrace, comparing internals and performance overhead. \r\nThe talk also covers low-overhead, statically defined tracepoints and counter-based sampling via perf_event. Finally, we introduce BPF as a programmable execution layer that reuses and enhances tracing sources for flexible and efficient kernel observation.",{"title":978,"describe":979,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FFKFEQJ",{"id":984,"room":985,"start":499,"end":500,"language":96,"track":986,"speakers":988,"zh":996,"en":999,"tags":1002,"uri":1003},"FSQWXX",{"en":684,"zh-hant":10},{"id":686,"name":987},{"en":688,"zh-hant":689},[989],{"id":990,"avatar":991,"zh":992,"en":995},"9CGX78","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002F9CGX78_1i9tujE.webp",{"name":993,"bio":994},"Ryan Hsieh 謝文豪","我是一位對開源充滿熱忱的 TPM。目前我在國泰金控與夥伴一同推動 OSPO，在 OSRB 治理與 DevRel 社群經營之間建立橋樑。\r\n\r\n目前正致力於開拓一條專屬於 PM 的開源之路，近期深耕研究 OpenSSF Scorecard, OSPS Baseline ,S2C2F 的相關國際專案。除了參與 2026 OSPOlogy Asia 委員會與 TODO Group，我也熱衷於將大型企業的實務帶回社群。\r\n\r\n說到底，我熱愛我所做的事，也相信開源帶來更好的世界。此外，我也是個網球愛好者🎾\r\n\r\n\r\nI’m a Technical Project Manager with a deep passion for open-source. At Cathay Financial Holdings, I work alongside squad to drive OSPO initiatives, bridging the gap between OSRB governance and DevRel community engagement.\r\n\r\nI am currently on a mission to carve out a unique path for PMs in the open-source ecosystem. My recent focus has been on the practical implementation of international projects such as the OpenSSF Scorecard, OSPS Baseline, and S2C2F. Beyond my daily work, I serve on the OSPOlogy Asia 2026 Program Committee and contribute to the TODO Group, where I advocate for bringing real-world enterprise practices back to the community.\r\n\r\nIn the end, I love what I do and firmly believe that open collaboration builds a better world. Also, I’m a huge tennis lover! 🎾",{"name":993,"bio":994},{"title":997,"describe":998,"type":30},"企業開源治理的守門人：從 OpenSSF Scorecard 到 S2C2F 的實踐之路","隨著開源軟體成為現代企業開發的基石，軟體供應鏈安全已不再是選擇題。實際上，企業在推動開源治理時，往往面臨「標準模糊」與「開發者反彈」兩大挑戰。\r\n\r\n本場演講將分享在企業內部推動開源治理借鏡國際社群 **OpenSSF** 經驗。我們將從 **OpenSSF Scorecard** 的自動化評分機制出發，探討如何量化開源專案的安全風險；接著引入 **S2C2F (Secure Supply Chain Consumption Framework) 與 OSPO Baseline**，建構一套完整的企業開源治理框架。",{"title":1000,"describe":1001,"type":30},"Scaling Enterprise OSS Governance with OpenSSF Tools","Software supply chain security is critical, yet enterprises often struggle with implementation. This session explores the practical application of OpenSSF Scorecard and S2C2F in a corporate environment. We will cover quantifying open-source risk, navigating license compliance, and adopting phased rollout strategies to balance security with developer velocity. Gain actionable insights on bridging the gap between theoretical frameworks and real-world governance.",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FFSQWXX",{"id":1005,"room":1006,"start":709,"end":68,"language":70,"track":1008,"speakers":1010,"zh":1018,"en":1021,"tags":1022,"uri":1023},"GAS93K",{"en":1007,"zh-hant":10},"TR512",{"id":901,"name":1009},{"en":903,"zh-hant":904},[1011],{"id":1012,"avatar":1013,"zh":1014,"en":1017},"ADVTWE","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FADVTWE_F4rjjk1.webp",{"name":1015,"bio":1016},"Sungjoon Moon","Free and Open Source Software contributor\r\nCo-founder and Director of [FOSS for All](https:\u002F\u002Ffossforall.org\u002Fen)\r\n\r\nInterested in RISC-V, wayland, CJK.",{"name":1015,"bio":1016},{"title":1019,"describe":1020,"type":30},"crossdev-stages, painless tool for building cross-compiler environment and board images for RISC-V and other architectures","Bringing Linux up on a new RISC-V board usually means assembling a vendor BSP and chasing toolchain mismatches by hand.\r\n\r\n**crossdev-stages** is a Rust CLI (and also a library) that wraps the whole flow in one declarative pipeline.\r\n\r\nEach board is a small set of config files in `boards\u002F\u003Cname>\u002F` (arch, CFLAGS, kernel \u002F OpenSBI \u002F U-Boot \u002F firmware repos & tags, services), with optional `pre\u002Fpost\u002Foverride` shell hooks per build step.\r\nThe tool drives Gentoo's `crossdev` to build a cross-toolchain, fetches pinned sources, cross-emerges the rootfs with whatever Portage packages and USE flags the board declares, and assembles a bootable image.\r\n\r\nIt runs inside [hakoniwa](https:\u002F\u002Fgithub.com\u002Fsouk4711\u002Fhakoniwa) (箱庭): an unprivileged user-namespace sandbox, so no `sudo` or Docker daemon.\r\n\r\nCross-toolchains and binary packages are keyed by `(CHOST, canonical CFLAGS)`; CFLAGS go through [sokgi](https:\u002F\u002Fgithub.com\u002FOctopusET\u002Fsokgi) first, which normalizes semantically equivalent flag strings (token order, last-wins overrides) to the same hash.\r\n\r\nSo one toolchain build per hardware profile, shared across compatible boards. Each image build also writes a `build.lock.toml` pinning stage3, source commits, kernel\u002FU-Boot config sha256, and CFLAGS. `crossdev-stages update` diffs it against upstream HEAD when you want to refresh.",{"title":1019,"describe":1020,"type":30},[402],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FGAS93K",{"id":1025,"room":1026,"start":500,"end":11,"language":96,"track":1027,"speakers":1029,"zh":1037,"en":1040,"tags":1043,"uri":1044},"GC99EF",{"en":167,"zh-hant":10},{"id":169,"name":1028},{"en":171,"zh-hant":172},[1030],{"id":1031,"avatar":1032,"zh":1033,"en":1036},"EDGUYA","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FEDGUYA_wbAgePW.webp",{"name":1034,"bio":1035},"Tsai,Ming-Han","是個對複雜的事物都很喜歡且很有自己想法的一個人。\r\n\r\nps 我在找工作 有需要都歡迎聯繫!\r\nas95630as@gmail.com",{"name":1034,"bio":1035},{"title":1038,"describe":1039,"type":30},"突破 FP64 限制：AdaptiveGEMM 透過 INT8 Tensor Core 在消費級 GPU 逼近 A100 效能","高精度矩陣乘法 (GEMM) 在科學與 AI 運算中是核心算子。然而，消費級 GPU (如 RTX 40 系列) 的 FP64 雙精度算力受硬體限制，成為許多開發者與量化交易的效能瓶頸。\r\n\r\n本議程將分享開源專案 AdaptiveGEMM：探討如何利用實作 Ozaki Scheme 將高精度運算降維拆解。我們將深入底層，利用 CUDA PTX 與 mma.sync 指令極限壓榨 INT8 Tensor Core，解決 Shared Memory Bank Conflict，成功在 RTX 4060 上大幅提升效能，達到逼近 A100 等級的 FLOPS 表現。\r\n\r\n【難易度：進階】適合尋求突破硬體極限的 HPC 與 AI 底層架構工程師。\r\n【先備知識】建議具備基礎 C++ 能力，了解 GEMM 運作原理，並對 GPU 記憶體架構與 Tensor Core 有初步認識。",{"title":1041,"describe":1042,"type":30},"Breaking FP64 Limits: AdaptiveGEMM Achieves Near-A100 Performance on RTX 4060 using INT8 Tensor Cores","High-precision GEMM is vital for scientific computing and AI. However, consumer GPUs (like the RTX 40 series) are hindered by hardware limitations in FP64 performance.\r\n\r\nThis session introduces \"AdaptiveGEMM,\" an open-source project employing the Ozaki Scheme to algorithmically decompose high-precision operations. We will dive into CUDA architecture, utilizing PTX and mma.sync instructions to maximize INT8 Tensor Core utilization while resolving Shared Memory Bank Conflicts. Discover how this project bypasses hardware restrictions to achieve near A100-level FLOPS on an RTX 4060.\r\n\r\nDesigned for advanced HPC engineers, this talk requires basic C++ proficiency and an understanding of GPU memory hierarchy.",[402],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FGC99EF",{"id":1046,"room":1047,"start":238,"end":239,"language":96,"track":1048,"speakers":1050,"zh":1054,"en":1057,"tags":1058,"uri":1059},"GFKLC8",{"en":263,"zh-hant":10},{"id":44,"name":1049},{"en":46,"zh-hant":47},[1051],{"id":268,"avatar":269,"zh":1052,"en":1053},{"name":47,"bio":271},{"name":47,"bio":271},{"title":1055,"describe":1056,"type":30},"威脅模型與 Metadata 入門：認識你的對手，以及為什麼匿名工具要開源才可信","挑工具之前，先回答三個問題：你要保護什麼、要防的對手是誰、願意付出多少成本。這場用威脅模型把這三個問題搭成判斷框架，讓你在遇到新工具、新威脅時知道如何對號入座。這個框架裡有一條和開源直接相關。當你的對手有能力檢查你的裝置與通訊時，你用的工具能不能被獨立稽核就攸關安全，這也是 Tor Browser、Tails 採用開放原始碼與可重現建置（reproducible build，任何人都能從原始碼編出位元相同的程式、驗證沒被動過手腳）的原因。接著談 metadata（那些沒被加密的連線、時間、聯絡對象等紀錄），就算通話內容沒被監聽，對手仍可能從通聯紀錄知道你在什麼時間聯繫了誰、聯繫多久。對記者與公民團體來說，這類紀錄往往足以反推出消息來源與行動網絡。",{"title":1055,"describe":1056,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FGFKLC8",{"id":1061,"room":1062,"start":239,"end":408,"language":70,"track":1064,"speakers":1069,"zh":1084,"en":1087,"tags":1088,"uri":1089},"GJV9WT",{"en":1063,"zh-hant":10},"TR410",{"id":1065,"name":1066},528,{"en":1067,"zh-hant":1068},"OSPN (Open Source People Network) Japan","OSPN（Open Source People Network）日本",[1070,1077],{"id":1071,"avatar":1072,"zh":1073,"en":1076},"X7CDAJ","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FX7CDAJ_yqMPGfw.jpg",{"name":1074,"bio":1075},"K.Ito","K. Ito originally worked as a healthcare professional. Following an organizational restructuring, he was reassigned to the IT department.\r\n\r\nTasked by the organization's president with reducing costs and improving operational efficiency, he became involved in the adoption and promotion of open source software. This experience led him to discover the value of open source communities and technologies.\r\n\r\nToday, his organization relies on Nextcloud for collaboration and LibreOffice for office productivity, both of which play an important role in its daily operations.",{"name":1074,"bio":1075},{"id":1078,"avatar":1079,"zh":1080,"en":1083},"NL7NXD","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FNL7NXD_nT6yXc7.webp",{"name":1081,"bio":1082},"Yuuichi Kojima","Yuuichi Kojima is a 45-year-old healthcare professional based in Tokyo, Japan. In his daily work, he supports and works closely with many people.\r\n\r\nHe is not an IT professional and never felt comfortable joining conversations about technology. However, when LibreOffice was introduced in his workplace, a colleague who led the project introduced him to the world of open source software (OSS). Since then, he has gradually become involved with OSS communities.\r\n\r\nToday, he uses LibreOffice and Nextcloud as part of his daily work. After nearly five years of using them, he would like to share his experiences with the audience in this session.",{"name":1081,"bio":1082},{"title":1085,"describe":1086,"type":30},"How My Daily Life Changed After Discovering LibreOffice","A person who had no interest in IT started using LibreOffice after it was introduced by a colleague at work. This simple change became the first step into the world of open source software (OSS).\r\n\r\nThe colleague who introduced LibreOffice was also active in OSS communities. At first, LibreOffice was just another tool for daily work. However, through using it, he learned about the community behind it. By meeting new people and having new experiences, his world gradually became larger.\r\n\r\nIn this talk, we will look back on how a person with no interest in OSS became involved in an open source community. We will also share the story from the viewpoint of the colleague who helped him take that first step.\r\n\r\nWhy did a world that once seemed unrelated become interesting? What encouraged him to get involved? What did he learn through community activities, and how did his way of thinking and acting change? We will answer these questions through real experiences.\r\n\r\nThis story includes the worries and uncertainty of joining a community for the first time, lessons learned from mistakes, the satisfaction of working together with others, and the process of finding a role within a community. We will also talk about how experiences gained through OSS activities have influenced work and daily life.\r\n\r\nWe hope this talk will be useful for people who are interested in OSS but hesitate to join, for those who want to take a first step into a new community, and for those who think OSS has nothing to do with them.",{"title":1085,"describe":1086,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FGJV9WT",{"id":1091,"room":1092,"start":605,"end":476,"language":70,"track":1093,"speakers":1095,"zh":1103,"en":1106,"tags":1107,"uri":1108},"GMJZEK",{"en":237,"zh-hant":10},{"id":241,"name":1094},{"en":243,"zh-hant":243},[1096],{"id":1097,"avatar":1098,"zh":1099,"en":1102},"7FZPBY","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002F7FZPBY_ib1jWfq.webp",{"name":1100,"bio":1101},"Georgy Asfur","I'm Georgy (ゲオルギ), and I build backend applications written in Ruby (and sometimes in Go). Currently, I'm based in Osaka, Japan, and work at Evil Martians (https:\u002F\u002Fevilmartians.com\u002F), which puts a lot of effort into OSS. I occasionally contribute to OSS projects, but mainly I co-maintain Stoplight (https:\u002F\u002Fgithub.com\u002Fbolshakov\u002Fstoplight\u002F) - an OSS implementation of the circuit breaker pattern in Ruby.\r\nBesides writing code, I'm a huge coffee junkie, so I'll appreciate any coffee shop recommendations in Taipei.",{"name":1100,"bio":1101},{"title":1104,"describe":1105,"type":30},"Stoplight: Circuit Breakers for Ruby in the AI Era","A slow third-party API can take down your entire Rails app. One service starts timing out, requests pile up waiting for it, threads block, memory fills, and suddenly nothing responds - not even the parts of your app that have nothing to do with the failing service. This is a cascading failure.\r\n\r\nUntil recently, this was a problem for large engineering teams running dozens of services. But in 2026, even a small startup is calling out to LLMs, OCR engines, and other third-party APIs. Every one of those calls is a potential domino. Circuit breakers belong in any Ruby app shipping AI features.\r\n\r\nThis talk introduces the circuit breaker pattern through Stoplight, an MIT-licensed Ruby gem, I co-maintain with Tema Bolshakov. I'll start by unpacking how a single slow dependency cascades into a full outage, then walk through how Stoplight prevents it - and why its traffic-light metaphor (green\u002Fyellow\u002Fred) is more intuitive than the classical open\u002Fclosed\u002Fhalf-open vocabulary.\r\n\r\nThe second half peels back the curtain on how we made Stoplight fast. Coordinating circuit breaker state across multiple Ruby processes means hitting Redis on every call, and a naive implementation incurs several round-trips per check. We'll look at how we collapsed those into single atomic Lua scripts - a technique that improved Stoplight's performance substantially and applies to anyone building Ruby tooling on top of Redis.\r\n\r\n**Outline (30 min):**\r\n- (5 min) Cascading failures in the AI era\r\n- (6 min) The circuit breaker pattern and the traffic-light reframe\r\n- (6 min) Stoplight in practice: defaults, fallbacks, Rails integration\r\n- (8 min) Gem internals: improving performance with Lua scripts\r\n- (5 min) Stoplight admin panel, takeaways, Q&A\r\n\r\n**Target audience**: Ruby and Rails developers building apps with external or AI service dependencies. However, the concepts from this talk translate to any language.\r\n\r\n**Background**: Basic Ruby and a rough idea of what Redis is. No prior exposure to the circuit breaker pattern or the concept of cascading failure is needed.\r\n\r\n**Links:**\r\n- **Stoplight gem**: https:\u002F\u002Fgithub.com\u002Fbolshakov\u002Fstoplight\r\n- **Companion article**: https:\u002F\u002Fevilmartians.com\u002Fchronicles\u002Fcircuit-breakers-and-ruby-in-2025-dont-break-your-launch",{"title":1104,"describe":1105,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FGMJZEK",{"id":1110,"room":1111,"start":1112,"end":454,"language":70,"track":1113,"speakers":1115,"zh":1123,"en":1126,"tags":1127,"uri":1128},"GQGLDL",{"en":899,"zh-hant":10},"2026-08-08T14:20:00+08:00",{"id":901,"name":1114},{"en":903,"zh-hant":904},[1116],{"id":1117,"avatar":1118,"zh":1119,"en":1122},"MB399N","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FMB399N_0gXueZ6.webp",{"name":1120,"bio":1121},"Aaditya Binod Yadav","Aaditya Binod Yadav is a full-stack engineer and open-source advocate based in Kathmandu, Nepal, currently pursuing Computer Engineering at Nepal College of Information and Technology. He has led production-grade platform development across SaaS, cloud infrastructure, and API systems, and is the founder of Flint Secure, a real-time fraud detection and device intelligence platform targeting Nepal's digital payment ecosystem. His technical interests span distributed systems, low-latency infrastructure, and fintech engineering\r\n\r\nBeyond engineering, Aaditya is an active open-source tech educator having delivered workshops and talk sessions at St. Xavier's College and other institutions across Kathmandu, reaching hundreds of students across web development and systems topics. He served as a PR Associate for UbuCon Asia 2025 and GNOME Nepal, is a contributor at NOSK (Nepal Open Source Klub), was a Technical Associate at Google Developers Group Kathmandu, and leads TEDxNCIT. He came to COSCUP through his involvement in open-source communities across Asia, where he continues to teach, contribute, and build.",{"name":1120,"bio":1121},{"title":1124,"describe":1125,"type":30},"Building fraud detection infrastructure for the unprotected: open-source fintech from Nepal","South Asia's digital payment ecosystem is growing fast but the fraud infrastructure protecting it hasn't kept up. Platforms serving millions of users often lack real-time device intelligence, have no way to detect SIM swap attacks without KYC dependency, and can't share signals across providers without exposing customer data. The threat model is also genuinely different here: attackers exploit device cycling, identity reuse across platforms, and the absence of cross-institution coordination in ways that Western fraud tooling simply isn't designed for.\r\n\r\nThis talk explores how you can address these gaps using open-source infrastructure: Go, NATS, PostgreSQL, Python and why releasing client SDKs openly matters as much as the system design itself. We'll cover phone-hash-based SIM swap detection, weighted real-time fraud scoring with cold-start constraints, device fingerprinting without invasive data collection, and the architecture of a scoring pipeline designed to run at payment speed.\r\n\r\nThese are lessons learned firsthand while building a fraud detection platform targeting Nepal's payment ecosystem the technical decisions, the dead ends, and what a working open-source approach actually looks like in practice.",{"title":1124,"describe":1125,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FGQGLDL",{"id":1130,"room":1131,"start":830,"end":1132,"language":96,"track":1133,"speakers":1135,"zh":1143,"en":1146,"tags":1149,"uri":1150},"HEV7GN",{"en":121,"zh-hant":10},"2026-08-08T16:20:00+08:00",{"id":733,"name":1134},{"en":735,"zh-hant":735},[1136],{"id":1137,"avatar":1138,"zh":1139,"en":1142},"ALHRTE","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FALHRTE_yZMX9ae.webp",{"name":1140,"bio":1141},"林源隆（Wilson）","近期關注 AI、勞動市場與數位轉型領域，研究興趣包含生成式 AI 對職能結構的影響、技能變遷、資料分析與科技應用。過去曾參與 AI、NLP、資料探勘與社會科學研究相關專案，且持續探索如何將大型語言模型、開源工具及自動化工作流應用於公共議題分析。",{"name":1140,"bio":1141},{"title":1144,"describe":1145,"type":30},"打造開源勞動觀測站：用 n8n 與 LLM 解析 AI 時代的職能轉變與就業衝擊","AI 時代下，勞動市場正面臨快速且深刻的職能轉變。然而求職網站的大量非結構化資料，以及分散於各處的政府開放資料，往往使一般公民與社會科學研究者難以即時掌握勞動市場變化。\r\n\r\n本議程將分享如何運用開源視覺化自動化工具 n8n，結合大型語言模型（Large Language Model, LLM），打造一套「勞動觀測站」。透過將傳統社會科學研究流程「Public Code」，將展示如何建立一套能夠自動爬梳、整理、分析並視覺化勞動市場趨勢的資料管線（Data Pipeline）。\r\n\r\n在此架構中，系統進一步建立自動化風險指標（Automation Risk Index），以量化 AI 對不同年資、職能與產業勞工所帶來的潛在衝擊。這套方法不僅助於即時追蹤各產業的技能漂移（Skill Drift）現象，也能降低公民參與數位政策研究與勞動市場分析的技術門檻，實踐科技賦能社群與開放知識共享的理念。",{"title":1147,"describe":1148,"type":30},"Building an Open-Source Labor Observatory: Using n8n and LLMs to Track Skill Drift and AI Employment Risks","How is AI reshaping jobs, skills, and career risks?\r\n\r\nThis session presents an open-source Labor Observatory built with n8n and LLMs. It shows how automated data pipelines can collect, analyze, and visualize labor market trends from job platforms and open data sources.\r\n\r\nBy tracking skill drift and exploring automation risks across industries and career stages, this project aims to make labor market research more open, accessible, and civic-oriented.",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FHEV7GN",{"id":1152,"room":1153,"start":95,"end":1155,"language":328,"track":1156,"speakers":1158,"zh":1166,"en":1169,"tags":1170,"uri":1171},"HNFEJJ",{"en":1154,"zh-hant":10},"RB101","2026-08-08T15:30:00+08:00",{"id":330,"name":1157},{"en":332,"zh-hant":332},[1159],{"id":1160,"avatar":1161,"zh":1162,"en":1165},"L39WHT","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FL39WHT_8q5yEYk.webp",{"name":1163,"bio":1164},"Soumyadeep Ghosh","Soumyadeep Ghosh is an Ubuntu Member and contributor. They have been contributing around the tooling and ecosystem of snapd for the last 4 years. An active contributer and a prolific snapcrafter, also a part of the Snapcrafters community and actively maintaining 30+ snaps.",{"name":1163,"bio":1164},{"title":1167,"describe":1168,"type":30},"snappy-debug: demystify and extend snapd","In this talk, I'll start by discussing apparmor, cgroup, seccomp etc and talk briefly about how these all are used to create a confinement for snapped apps. Then, I'll deep dive into the world of debugging confinement issues. I'll showcase and give a small hands-on on the snappy-debug tool and show how one can use it, in different scenarios, like in the Ubuntu Core or the classic Ubuntu distro and end the talk by showing a small example, on how one can actually extend various interfaces of snapd.\r\n\r\nMore info: https:\u002F\u002Fevents.canonical.com\u002Fevent\u002F146\u002Fcontributions\u002F938\u002F",{"title":1167,"describe":1168,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FHNFEJJ",{"id":1173,"room":1174,"start":1175,"end":1176,"language":96,"track":1177,"speakers":1179,"zh":1187,"en":1190,"tags":1191,"uri":1192},"JCR78M",{"en":922,"zh-hant":10},"2026-08-08T15:15:00+08:00","2026-08-08T15:45:00+08:00",{"id":924,"name":1178},{"en":926,"zh-hant":927},[1180],{"id":1181,"avatar":1182,"zh":1183,"en":1186},"9YK8CR","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002F9YK8CR_3JYJ4Ql.webp",{"name":1184,"bio":1185},"kc","陳貴成 KC Chen\r\n\r\n現為 OCF 開放文化基金會董事，自由軟體創業者，為中華民國軟體自由協會 (SLAT) 之發起人之一並曾擔任多屆\u002F理監事、OSSACC 營運長，並為 ezgo 光碟、COSCUP 等自由軟體專案或計畫早期發起人之一。2005 年創立智新資通股份有限公司並以自由軟體服務為企業核心發展方向迄今，並於 2024 年與友人合資成立源智匯股份有限公司，以 AI 落地實現運用為主要核心業務。",{"name":1184,"bio":1185},{"title":1188,"describe":1189,"type":30},"堅持開源之道的公司怎麼賺錢長大活了2X年？","- 智新資通怎麼用開源軟體做生意?\r\n- 為什麼選擇用開源軟體? 好處和壞處? 會特別在意是哪種 release license 嗎?\r\n- 客戶在意是開源軟體嗎?\r\n- 既然是 open source，客戶為什麼會花錢買?\r\n- 用開源軟體要怎麼競爭對手有差異?",{"title":1188,"describe":1189,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FJCR78M",{"id":1194,"room":1195,"start":604,"end":605,"language":42,"track":1196,"speakers":1198,"zh":1206,"en":1209,"tags":1210,"uri":1211},"JFMB7M",{"en":899,"zh-hant":10},{"id":901,"name":1197},{"en":903,"zh-hant":904},[1199],{"id":1200,"avatar":1201,"zh":1202,"en":1205},"XN8RE9","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FXN8RE9_TagSBf8.webp",{"name":1203,"bio":1204},"Louis Yoong 翁欣瑜","Louis Yoong is a Senior Software Developer from Malaysia currently working at Tetra Tech RPS Energy, building web and mobile solutions related to seismic survey optimisation and ocean monitoring. She also serves as a part-time UI\u002FUX lecturer, sharing knowledge in design and modern web development with students.\r\n\r\nHer work focuses on developing interactive geospatial platforms and data visualization systems for offshore development, seismic operations, and environmental monitoring. Louis is also a Cypress Ambassador and former GDG George Town Lead, actively contributing as a tech speaker across Southeast Asia, where she shares insights on web development, interactive maps, and geospatial technologies.",{"name":1203,"bio":1204},{"title":1207,"describe":1208,"type":30},"Building AI-Powered Interactive Maps with Open Data 如何使用開放資料建構 AI 驅動的互動式地圖","This session will shows how to build interactive 2D and 3D maps using open datasets and open geospatial formats. Through real-world website examples, I’ll demonstrate how users can explore weather, location, and environmental data in an interactive and intuitive way.\r\n\r\nAlso covers practical approaches to improve scalability, such as better data organization and map layer handling. While Mapbox and Google Maps Platform are used as examples, including customizing map styles with google cloud based tools the focus is on reusable ideas and patterns that work across different mapping libraries and technology stacks.",{"title":1207,"describe":1208,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FJFMB7M",{"id":1213,"room":1214,"start":499,"end":500,"language":96,"track":1215,"speakers":1217,"zh":1225,"en":1228,"tags":1231,"uri":1232},"JK7TYL",{"en":121,"zh-hant":10},{"id":125,"name":1216},{"en":127,"zh-hant":128},[1218],{"id":1219,"avatar":1220,"zh":1221,"en":1224},"GMSPGP","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FGMSPGP_aviY3YG.webp",{"name":1222,"bio":1223},"Jing-Tian Sung","試圖在冰冷的模型參數中，找回技術應有人文溫度的工程師一枚。不只是在校正誤差，更想修復的是科技與社會之間的信任鏈。",{"name":1222,"bio":1223},{"title":1226,"describe":1227,"type":30},"邁向可信任 AI：開發者必備的 Agents 偏見檢測與緩解框架","隨著大型語言模型（LLM）驅動的 Agents 逐漸走入自動化工作流，從履歷篩選、客服對話到商業決策等，Agent 的自主性在提升效率的同時，也悄然埋下了不同的偏見與歧視的風險。由於 Agent 具備推理與調用外部工具的能力，其偏見往往比單純的 LLM 更加隱蔽且具備行動力，如何有效地評估這些智慧體，成為構建可信 AI 的關鍵挑戰。\r\n\r\n在此次將分享深入探討一套開源的 Agents 偏見檢測工具。我們將從技術角度解構 Agent 產生偏見的根源，並介紹如何透過自訂的知識庫與資料集來建立自動化基準測試，對 Agent 在特定情境下的是否有偏見的行為進行量化評估。\r\n\r\n本次分享的核心內容包括：\r\n- 偏見識別維度： 定義 Agent 在應用情境中的偏見樣態與基準調整。\r\n- 評測工具架構： 介紹如何利用開源框架建構自動化測試腳本，模擬多樣化用戶輸入以誘發潛在偏見。\r\n- 實戰案例分析： 展示該工具在實際場景（如徵才 Agent 或決策輔助 Agent）中的檢測成果。\r\n- 緩解策略： 探討如何透過 Prompt Engineering 或約束層（Guardrails）在開發階段降低偏見風險。\r\n\r\n我們希望透過開源工具，賦予開發者主動監測與修正 Agent 行為的能力。這不僅是一場技術分享，更是呼籲開源社群共同參與，為 AI 時代建立一道公平與透明的防火牆。",{"title":1229,"describe":1230,"type":30},"Towards Trustworthy AI: Essential Open-Source Tools for Agent Bias Detection and Mitigation","As LLM-driven Autonomous Agents become increasingly integrated into automated workflows—ranging from resume screening and customer service to strategic business decisions—their autonomy brings both efficiency and the hidden risk of social bias and discrimination. Because Agents possess the ability to reason and invoke external tools, their biases are often more covert and \"actionable\" than those of standard LLMs. Evaluating these agents effectively has become a critical challenge in building Trustworthy AI.\r\n\r\nIn this session, we will dive deep into an open-source bias detection toolkit for Agents. We will deconstruct the technical roots of agentic bias and introduce how to build automated benchmarks using custom knowledge bases and datasets to quantitatively evaluate biased behaviors in specific scenarios.\r\n\r\nKey highlights of this session include:\r\n\r\nDimensions of Bias Identification: Defining bias patterns and benchmark adjustments within specific application contexts.\r\n\r\nEvaluation Framework Architecture: How to leverage open-source frameworks to build automated testing scripts that simulate diverse user inputs to trigger potential biases.\r\n\r\nReal-world Case Studies: Showcasing detection results in practical scenarios, such as recruitment agents or decision-support systems.\r\n\r\nMitigation Strategies: Exploring ways to reduce bias risks during the development phase through Prompt Engineering and Guardrails.\r\n\r\nThrough these open-source tools, we aim to empower developers to proactively monitor and rectify Agent behavior. This is more than a technical showcase; it is a call to the open-source community to collaborate in building a firewall of fairness and transparency for the AI era.",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FJK7TYL",{"id":1234,"room":1235,"start":95,"end":830,"language":96,"track":1236,"speakers":1238,"zh":1242,"en":1245,"tags":1248,"uri":1249},"JPRTNQ",{"en":9,"zh-hant":10},{"id":15,"name":1237},{"en":17,"zh-hant":18},[1239],{"id":459,"avatar":460,"zh":1240,"en":1241},{"name":462,"bio":463},{"name":462,"bio":463},{"title":1243,"describe":1244,"type":59},"MediaWiki暨維基媒體專案漢語族語言峰會","本議程集合各漢語族語言以及各臺灣本土語言，共同討論在MediaWiki及維基媒體專案下如何強化各語言的發展。",{"title":1246,"describe":1247,"type":59},"MediaWiki and Wikimedia Projects Sinitic Languages Summit","This session is about discussing the development of MediaWiki and Wikimedia projects in Sinitic languages and other Taiwan native languages.",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FJPRTNQ",{"id":1251,"room":1252,"start":1254,"end":1255,"language":1256,"track":1257,"speakers":1262,"zh":1269,"en":1272,"tags":1273,"uri":1274},"JQWCV3",{"en":1253,"zh-hant":10},"RB105","2026-08-08T16:35:00+08:00","2026-08-08T16:45:00+08:00","其他",{"id":1258,"name":1259},512,{"en":1260,"zh-hant":1261},"Main Session Track","主議程軌 - Main Session Track",[1263],{"id":1264,"avatar":1265,"zh":1266,"en":1268},"9NRTBF","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002F52304773369_70b54c758b_c_1_KsktUqG.jpeg",{"name":1267,"bio":10},"COSCUP Team",{"name":1267,"bio":10},{"title":1270,"describe":1271,"type":30},"Recap Day 1","Closing Day 1",{"title":1270,"describe":1271,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FJQWCV3",{"id":1276,"room":1277,"start":1278,"end":499,"language":328,"track":1279,"speakers":1281,"zh":1296,"en":1299,"tags":1300,"uri":1301},"JRCCHW",{"en":167,"zh-hant":10},"2026-08-08T11:00:00+08:00",{"id":169,"name":1280},{"en":171,"zh-hant":172},[1282,1289],{"id":1283,"avatar":1284,"zh":1285,"en":1288},"B7RWEC","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FB7RWEC_gGdx6YS.webp",{"name":1286,"bio":1287},"Hrittik Roy","Hrittik is a Platform Advocate at Loft Labs and a CNCF Ambassador, with expertise in cloud native technologies and open source communities. He has contributed extensively to developer advocacy, technical writing, and community engagement. Hrittik has been a featured speaker at events such as Kubernetes Community Days, Open Source Summits, and more, and has served as a Program Committee member for several KubeCons and CloudNativeCons.",{"name":1286,"bio":1287},{"id":1290,"avatar":1291,"zh":1292,"en":1295},"P3JZJE","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FDisplay_Picture_rAx4fAJ.jpeg",{"name":1293,"bio":1294},"Parth Goswami","Parth Goswami is an open source enthusiast and mostly works with containers and networking. He loves contributing to open source by hosting meetups and interacting with the community.",{"name":1293,"bio":1294},{"title":1297,"describe":1298,"type":30},"YAML is the New Dockerfile: Building AI Agent Systems with Docker cagent","The industry is rapidly moving from single LLM applications toward systems composed of multiple agents, tools, memory layers, and workflows. But most discussions around AI agents focus heavily on frameworks and demos, while avoiding the larger systems question:\r\n\r\nWhat does agent infrastructure actually look like?\r\n\r\nThis talk takes a cloud native and systems-oriented view of AI agents. We’ll explore how ideas from distributed systems, containers, orchestration, and platform engineering are beginning to shape the next generation of AI tooling.\r\n\r\nTopics include:\r\n\r\n- Why AI agents resemble distributed systems more than chatbots\r\n- The emerging role of protocols like MCP\r\n- Declarative workflows versus imperative orchestration\r\n- Reproducibility and portability challenges in AI systems\r\n- Why infrastructure concepts like OCI artifacts and YAML workflows are reappearing in AI tooling\r\n\r\nUsing Docker cagent as a concrete example rather than the centerpiece, we’ll examine how modern agent runtimes are evolving toward infrastructure-style abstractions.",{"title":1297,"describe":1298,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FJRCCHW",{"id":1303,"room":1304,"start":238,"end":709,"language":96,"track":1305,"speakers":1307,"zh":1315,"en":1318,"tags":1321,"uri":1322},"JUXUWC",{"en":922,"zh-hant":10},{"id":946,"name":1306},{"en":948,"zh-hant":949},[1308],{"id":1309,"avatar":1310,"zh":1311,"en":1314},"CUAEJP","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002Fivan-tu_OssmmN6.jpg",{"name":1312,"bio":1313},"Ivan Tu","杜修文是 Oracle 的資深 MySQL 解決方案工程經理，領導東亞區域。\r\nMySQL 解決方案工程團隊致力於推廣 MySQL 技術，並協助關鍵企業用戶採用 MySQL。在社群參與方面，Ivan 是台灣 MySQL 使用者小組的組織者，該社群活躍，擁有超過 6,200 名參與者。\r\n欲了解更多關於Ivan Tu的背景資訊，請參閱他的LinkedIn頁面 https:\u002F\u002Fwww.linkedin.com",{"name":1312,"bio":1313},{"title":1316,"describe":1317,"type":30},"綜覽新一代MySQL長期支持版- MySQL 9.7","MySQL自前年四月將版本發行方式改為長期支持版(LTS)和創新版併行,並且推出了它的第一個LTS版,經過兩年MySQL 9.x創新版的累積新技和新功能,在今年4月推出了它的第二個LTS版 - MySQL 9.7,這個版本將享有8年的支援週期,這段期間9.7將持續提供更新版,而不會加新功能以保證應用系統在更新版之間的相容性,這追求系統穩定性的使用者而言,MySQL 9.7將是一個重要的里程碑-您可以安心的在8年內使用越來越穩定MySQL 9.7.\r\n在這個分享中我將為各位介紹MySQL 9.7帶來的最新功能,以及MySQL長期支持版的使用者能由9.7版得到什麼效益.",{"title":1319,"describe":1320,"type":30},"Introduction to MySQL 9.7","This session will introduce the latest features from MySQL 9.7 and the benefits that MySQL LTS offers to its users",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FJUXUWC",{"id":1324,"room":1325,"start":499,"end":193,"language":96,"track":1326,"speakers":1328,"zh":1336,"en":1339,"tags":1341,"uri":1342},"JXZPYZ",{"en":922,"zh-hant":10},{"id":946,"name":1327},{"en":948,"zh-hant":949},[1329],{"id":1330,"avatar":1331,"zh":1332,"en":1335},"YUJUNM","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FYUJUNM_WoMrgig.webp",{"name":1333,"bio":1334},"Hank Tom","軟體開發資歷 30 年，歷經多波技術變革。曾任程式設計師、軟體部經理、技術顧問與總監，主導過數十個業界專案，涵蓋即時聊天、企業級 SIP 通訊、社群分享等 App 應用系統。長期擔任多所學校課程講師，以及軟體公司與新創團隊的企業內訓講師。為全球第 18 位取得 Google Certified Android Developer 認證者，「綠豆湯」技術部落格發起人，持續發表 Android、Java、Firebase 等領域技術文章。",{"name":1333,"bio":1334},{"title":1337,"describe":1338,"type":59},"從零打造一個 MySQL-based RAG 系統：VECTOR 型別實戰、工程取捨與 pgvector 對照","當大家談到 RAG（Retrieval-Augmented Generation），第一個想到的儲存層往往是 Pinecone、Qdrant、Milvus，或是 PostgreSQL + pgvector。但是在絕大多數企業既有的 stack 裡，MySQL 才是那個「明明就在那裡、卻幾乎沒有人在 RAG 選型討論中提到」的角色。\r\n    MySQL 9.0 已經正式加入 VECTOR 資料型別。那麼問題來了：在不另外搬一座資料庫的前提下，到底能不能用 MySQL 把一個可運作的 RAG 系統做出來？做得起來、又能撐多久？\r\n    本場次以一個從零打造、完整開源的 demo 專案為主軸，帶聽眾一步步在 MySQL 9.x 上構建 RAG 系統：從 schema 設計、embedding 寫入、Top-K 相似度查詢，到串接 LLM 完成問答。途中會深入 VECTOR 型別的內部儲存方式、可用函式，以及目前最關鍵的限制——社群版尚未提供原生 ANN 索引。\r\n    接著，我們把同一份資料、同一組查詢搬到 PostgreSQL pgvector 上做對照組 benchmark，誠實呈現兩者在 latency、recall、開發體驗、維運成本上的差距。最後提出一個務實的選型決策框架：什麼情境下「在 MySQL 上做 RAG」是合理的工程選擇，什麼情境下應該果斷換工具。\r\n    所有 demo 程式碼、SQL schema、benchmark 腳本與投影片都會以 Apache-2.0 \u002F CC BY-SA 授權公開於 GitHub，現場聽眾可以即時在自己的環境重現。",{"title":1340,"describe":1338,"type":59},"Building a RAG System on MySQL from Scratch — VECTOR Type in Practice, Engineering Trade-offs, and a pgvector Reference",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FJXZPYZ",{"id":1344,"room":1345,"start":238,"end":239,"language":70,"track":1346,"speakers":1348,"zh":1356,"en":1359,"tags":1360,"uri":1361},"K79YPU",{"en":522,"zh-hant":10},{"id":524,"name":1347},{"en":526,"zh-hant":10},[1349],{"id":1350,"avatar":1351,"zh":1352,"en":1355},"FPNLZG","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FFPNLZG_R9H7FTd.jpg",{"name":1353,"bio":1354},"Christian Walter","Changing the world - with 9elements.",{"name":1353,"bio":1354},{"title":1357,"describe":1358,"type":30},"Open-Source Firmware on Framework Laptops","This talk is about the current status of Open-Source Firmware on Framework Laptops. We will dive through the portfolio of the very popular notebooks from Framework and dig into the current status of Open-Source Firmware on them. Which one's could be potential targets? We will also showcase a PoC running coreboot on one of the AMD Framework Laptops.",{"title":1357,"describe":1358,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FK79YPU",{"id":1363,"room":1364,"start":1365,"end":604,"language":96,"track":1366,"speakers":1368,"zh":1376,"en":1379,"tags":1382,"uri":1383},"K7EGEF",{"en":93,"zh-hant":10},"2026-08-08T13:10:00+08:00",{"id":98,"name":1367},{"en":100,"zh-hant":101},[1369],{"id":1370,"avatar":1371,"zh":1372,"en":1375},"BGF3WA","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002Fsd4Sk5rd_400x400_KMmd1Zy.jpg",{"name":1373,"bio":1374},"Laurence Chen","IT 顧問、講者、作家。專注於協助企業導入 Modern Data Stack\r\n\r\n現職：\r\n* REPLWARE ( https:\u002F\u002Freplware.dev ) CEO\r\n* 台灣 Clojure 社群 ( https:\u002F\u002Fclojure.tw\u002F ) 線下活動主辦人\r\n* Taipei dbt Meetup 社群 ( https:\u002F\u002Fmedium.com\u002Fdbt-local-taiwan ) 線下活動主辦人\r\n\r\n著作：\r\n* 從試算表到資料平台：重構資料工程的技術與團隊 ( https:\u002F\u002Freplware.dev\u002Fmds )",{"name":1373,"bio":1374},{"title":1377,"describe":1378,"type":30},"Polylith：帶有函數式編程思考的軟體架構","- Monolith 架構的問題：難以測試局部變更。\r\n- Microservices 架構的痛點：當兩個服務需要同一段業務邏輯時，只能選：(1) 複製貼上、(2) 抽出共用 library (3) 抽出作為 service。\r\n- 那如果有一種軟體架構，它不將「程式碼邊界」和「部署邊界」綁定呢？\r\n- Polylith 讓你可以用接近於 Monolith 的開發體驗，同時擁有 Microservices 的部署彈性。",{"title":1380,"describe":1381,"type":30},"Polylith - a software architecture that applies functional thinking.","- Problem with Monolith architecture: Difficult to test isolated changes.\r\n- Pain point of Microservices architecture: When two services need the same business logic, you're forced to choose: (1) copy-paste, (2) extract a shared library, or (3) extract it as a service.\r\n- What if a software architecture that decouples \"code boundaries\" from \"deployment boundaries\"?\r\n- Polylith gives you a development experience close to a Monolith, while retaining the deployment flexibility of Microservices.",[402],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FK7EGEF",{"id":1385,"room":1386,"start":1176,"end":41,"language":96,"track":1387,"speakers":1389,"zh":1397,"en":1400,"tags":1401,"uri":1402},"K89MHJ",{"en":1253,"zh-hant":10},{"id":1258,"name":1388},{"en":1260,"zh-hant":1261},[1390],{"id":1391,"avatar":1392,"zh":1393,"en":1396},"WAYBCE","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FWAYBCE_jCoTLza.webp",{"name":1394,"bio":1395},"高天","高天，Databricks员工，Python核心开发者，pdb负责人，Apache Spark Committer，Bilibili up主。手写代码、人眼review非物质文化遗产传承人。",{"name":1394,"bio":1395},{"title":1398,"describe":1399,"type":30},"[Prime Session] 成为开源开发者——why & how","从2015年第一次把自己的项目放上github,距今已有十多年的时间。在这十年的时间里,我从一\r\n个会写一点C的程序员,成为了开源社区的一份子——Python核心开发者,Apache Spark\r\nCommitter。\r\n回首这段开源之路,会发现自己的进步和成果完全不是线性的。似乎永远是有好长一段时间的迷\r\n茫和原地徘徊,然后再忽然迎来一个里程碑。而徘徊和迷茫的时间,随着自己的成长越来越少,迎\r\n来里程碑的频率也越来越高。\r\n这让我觉得,或许有些经验是可以分享的。这些经验没办法让大家直接成为一个优秀的开源开发\r\n者,但是也许可以让迷茫的时间少一点,里程碑来的速度快一些。人是正反馈动物,我希望通过这\r\n次的分享,可以让更多的人及时迎来自己的正反馈,愉快并长久地成为开源社区的一份子。\r\n今天的这个演讲,我会试图回答两个我被问得最多的问题——为什么要做开源项目,和怎么做开\r\n源项目。我相信如果你有了这两个问题的答案,你就已经准备好成为一个开源项目开发者了!",{"title":1398,"describe":1399,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FK89MHJ",{"id":1404,"room":1405,"start":122,"end":216,"language":96,"track":1406,"speakers":1408,"zh":1416,"en":1419,"tags":1422,"uri":1423},"KCDCT7",{"en":167,"zh-hant":10},{"id":169,"name":1407},{"en":171,"zh-hant":172},[1409],{"id":1410,"avatar":1411,"zh":1412,"en":1415},"8KMRCD","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002F8KMRCD_sMRaa5t.webp",{"name":1413,"bio":1414},"Nero Un 阮智軒","來自澳門的開發者，現職 IBM 顧問，具備豐富的資料科學、資料工程與人工智慧領域實務經驗，曾參與醫療、金融及製造業等跨產業專案，累積多元的產業洞察、解決方案設計與系統開發經驗。 畢業於高雄醫學大學，並取得成功大學醫學資訊研究所碩士學位。曾於生醫新創公司擔任 R & D 及 TPM。 熱衷研究技術探索推動變革的可能，深信科技的力量能為世界帶來影響與改變。 A developer from Macao, currently serving as a Consultant at IBM, with practical expertise in data science, data engineering, and artificial intelligence. Graduated from Kaohsiung Medical University and holds a master’s degree in Medical Informatics from National Cheng Kung University. Previously served as an R & D engineer and TPM at a biomedical startup. Passionate about exploring technology to drive transformative change, and firmly believes in the power of technology to influence and reshape the world.",{"name":1413,"bio":1414},{"title":1417,"describe":1418,"type":30},"A Prompt Is Not All You Need：用開源工具打造 LLM 合成資料管線","在開源模型生態蓬勃發展的今天，開發者能輕易取得 Gemma4、Qwen3.5 等頂尖的開源大語言模型 (LLM)。然而，當我們想將這些技術落地到金融、醫療等高合規場景時，往往會撞上一面名為「資料隱私」的高牆。受限於個資 (PII) 與法規，團隊根本拿不到真實業務資料來進行 PoC、效果驗證或 RAG 系統的壓力測試。沒有可用資料進行驗證，再好的開源模型也難以推進到實際的業務中。\r\n\r\n面對「資料可取用性」的難題，合成資料 (Synthetic Data) 成為破局關鍵方法。\r\n\r\n許多團隊最初會嘗試手寫 Prompt 來生成測試資料。但當需求擴增至上千筆，且須同時滿足多樣性、邊界條件與邏輯一致性時，這種做法很快就會面臨品質失控、難以驗證與無法擴展的瓶頸。\r\n\r\n本議程將以 Nvidia Open Source 工具 Data Designer 為例、提供可重現的範例 pipeline，包含 synthetic PII \u002F domain QA seed schema、DAG config、validator 設計、LLM-as-a-Judge rubric，以及生成資料的品質檢查報告。所有範例資料與程式碼將以開源授權釋出，讓聽眾會後可以直接改造成自己的 RAG 測試資料或 PoC dataset。\r\n\r\n透過這個實戰案例，希望幫助開源開發者與企業 IT 團隊在缺乏真實資料的困境中，利用開源工具鏈「無中生有」打造高品質的測試資料集，讓 AI 落地的最後一哩路走得更穩健。",{"title":1420,"describe":1421,"type":30},"A Prompt Is Not All You Need: Building a LLM Synthetic Data Pipeline with Open-Source Tools","Today, as the open-source model ecosystem flourishes, developers can easily access top-tier open-source large language models (LLMs) like Gemma4 and Qwen3.5. However, when attempting to implement these technologies in highly regulated fields such as finance and healthcare, they often encounter a significant barrier called \"data privacy.\" Due to restrictions related to personally identifiable information (PII) and regulations, teams cannot access real business data for PoC, performance validation, or RAG system stress testing. Without usable data for validation, even the best open-source models struggle to be integrated into actual business scenarios.\r\n\r\nTo address the challenge of \"data accessibility,\" synthetic data becomes a key solution.\r\n\r\nMany teams initially try manual prompt engineering to generate test data. But when the volume increases to thousands of records, and requirements include diversity, boundary conditions, and logical consistency, this approach quickly faces quality control issues, validation difficulties, and scalability bottlenecks.\r\n\r\nThis agenda will take Nvidia's open-source tool Data Designer as an example, providing a reproducible pipeline that includes synthetic PII\u002Fdomain QA seed schema, DAG configuration, validator design, LLM-as-a-Judge rubric, and data quality inspection reports. All sample data and code will be released under open-source licenses, allowing participants to adapt them directly into their own RAG test data or PoC datasets after the session.\r\n\r\nThrough this practical case study, we hope to assist open-source developers and enterprise IT teams in creating high-quality test datasets \"out of thin air\" using open-source toolchains, helping the AI implementation journey be more robust at the final leg.",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FKCDCT7",{"id":1425,"room":1426,"start":605,"end":476,"language":328,"track":1427,"speakers":1429,"zh":1437,"en":1440,"tags":1441,"uri":1442},"KGA3TE",{"en":326,"zh-hant":10},{"id":330,"name":1428},{"en":332,"zh-hant":332},[1430],{"id":1431,"avatar":1432,"zh":1433,"en":1436},"H7QRKG","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FH7QRKG_wPRROmy.webp",{"name":1434,"bio":1435},"Bikalpa Dhakal","Bikalpa Dhakal is a software engineer and open-source enthusiast based in Pokhara, Nepal. He currently works in the Analytics and Processing team at Canonical, where he helps maintain the Charmed Apache Spark project and its repositories. Before joining Canonical, he worked for companies such as Logpoint. More recently at Canonical, he was involved in the technical integration between Charmed Apache Spark and Charmed Kubeflow. Outside of his professional work, Bikalpa's hobbies include cinema, traveling and aviation.",{"name":1434,"bio":1435},{"title":1438,"describe":1439,"type":30},"From Big Data to ML: Integrating Spark and Kubeflow with Juju","In modern data science, the \"Data Lakehouse\" is a powerful way to manage large-scale data. However, for many teams, the \"last mile\" of integration—connecting Machine Learning platforms like Kubeflow to big data processing engines like Apache Spark—remains a major technical challenge. In a shared Kubernetes cluster, managing security, user permissions, and connection strings across different user profiles often leads to manual errors and security risks.\r\n\r\nIn this session, we will explore an open-source solution to this problem using Juju and Charmed Operators. We will walk through a \"Multi-Integrator\" architecture where each Kubeflow profile receives its own secure connection to the data platform. Instead of manual configuration, we use automated integrators to inject Spark Service Accounts, configurations, and security secrets directly into user namespaces.\r\n\r\nBy using Charmed Spark and Charmed Kubeflow, we can create a modular system where data scientists can move from a simple notebook to a distributed Spark job without worrying about the underlying YAML files or credentials. We will discuss the logic of this design, how it handles multi-tenancy, and how it simplifies the path to a production-ready data lakehouse on Ubuntu.\r\n\r\nMore info: https:\u002F\u002Fevents.canonical.com\u002Fevent\u002F146\u002Fcontributions\u002F951\u002F",{"title":1438,"describe":1439,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FKGA3TE",{"id":1444,"room":1445,"start":1446,"end":1447,"language":96,"track":1448,"speakers":1450,"zh":1458,"en":1461,"tags":1464,"uri":1465},"KHMK3H",{"en":282,"zh-hant":10},"2026-08-08T13:45:00+08:00","2026-08-08T14:15:00+08:00",{"id":286,"name":1449},{"en":288,"zh-hant":288},[1451],{"id":1452,"avatar":1453,"zh":1454,"en":1457},"LPWBDH","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002Farale_aFEeuz9.jpg",{"name":1455,"bio":1456},"張友謙","今年已經是第N年與大家一起分享關於PostgreSQL的方方面面了.\r\n今年會有比較進階且與系統設計有關的議題.",{"name":1455,"bio":1456},{"title":1459,"describe":1460,"type":30},"數與量在資訊系統中的正確建模與應用","「數與量」是資訊系統設計中最基礎、卻也最容易被低估的核心議題。本議題將聚焦在如何正確處理「數與量」，並探討其在真實系統中的建模方法與常見陷阱。\r\n「數」是純數學概念，「量」則是帶有業務規則、單位、精度、範圍與文化差異的領域概念。從金融系統的四捨五入與精度累積、製造業的單位轉換與損耗計算、醫療系統的多語言文化處理，到電商訂單的複雜業務狀態，「數與量」的正確建模直接影響系統的穩定性、一致性與可維護性。\r\n議題將深入討論：\r\n\r\n「數」與「量」的本質差異，以及它對系統設計的根本影響\r\n在資料庫中正確處理「量」的實務方法（自訂型別、領域事件、物化視圖、觸發器等）\r\n常見的反模式與真實專案教訓（精度遺失、單位混亂、業務規則分散等）\r\n如何將「數與量」的思考融入領域驅動設計（DDD），讓 DDD 不再只是抽象的高階概念，而是能真正落地的系統思維\r\n\r\nDDD 在此僅作為輔助工具，用來幫助我們更清晰地定義與封裝「量」的業務規則，而非主導整個討論。真正的挑戰在於：如何在資料庫層與應用層之間取得平衡，讓「數與量」被正確、穩定且可維護地處理。\r\n本演講適合對資料庫設計、系統架構、領域建模有興趣的開發者與架構師。希望透過這個議題，幫助大家在實務中更務實、也更正確地面對資訊系統中最基礎卻關鍵的「數與量」問題。",{"title":1462,"describe":1463,"type":30},"Modeling \"Numbers and Quantities\" Correctly in Information Systems","\"Numbers and Quantities\" is one of the most fundamental yet underestimated topics in information system design. This talk focuses on the correct modeling and application of \"quantities\" in real-world systems, and how it should be properly integrated with Domain-Driven Design (DDD).\r\nWhile \"numbers\" are pure mathematical concepts, \"quantities\" carry business rules, units, precision, and cultural differences. From financial rounding and precision issues, manufacturing unit conversions and loss calculations, to medical multilingual processing and complex e-commerce order states, the correct modeling of quantities directly affects system stability, consistency, and maintainability.\r\nThis talk will explore:\r\n\r\nThe essential differences between \"numbers\" and \"quantities\"\r\nPractical database techniques for handling quantities (custom types, domain events, materialized views, triggers)\r\nCommon anti-patterns and lessons from real projects\r\nHow DDD should work with database design instead of treating it as a mere implementation detail\r\n\r\nThe goal is to help developers and architects think more systematically about one of the most critical foundations in information systems.",[402],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FKHMK3H",{"id":1467,"room":1468,"start":68,"end":69,"language":1469,"track":1470,"speakers":1472,"zh":1480,"en":1483,"tags":1484,"uri":1485},"KLUB3Q",{"en":1063,"zh-hant":10},"英語",{"id":1065,"name":1471},{"en":1067,"zh-hant":1068},[1473],{"id":1474,"avatar":1475,"zh":1476,"en":1479},"EWEN39","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002Feno250-250_I1LtFxH.jpg",{"name":1477,"bio":1478},"Shinji Enoki","Shinji Enoki (榎真治) is a member of the LibreOffice Japanese Team and The Document Foundation, where he also serves on the Membership Committee. He focuses on organizing events and community activities and occasionally contributes to QA efforts.\r\n\r\nHe is also involved in various open source and open data communities. His roles include staff member of the Wikimedians of Japan User Group, organizing member of the Nextcloud Japan community, and volunteer staff for the Japan UNIX Society and KANSAI OPEN SOURCE. He is also a casual mapper on OpenStreetMap, among other activities.\r\n\r\nHe is also a local organizing staff member for the Wikimedia ESEAP Summit 2027, to be held in Osaka.\r\n\r\nHe works as a freelancer and is developing a LibreOffice support business in Japan",{"name":1477,"bio":1478},{"title":1481,"describe":1482,"type":30},"From Events to Relationships: Building Connections Across Asian Open Source Communities","In Asian open source communities, conferences and regional events play an important role in creating personal connections across countries and projects. However, maintaining collaboration after events is often much more difficult than starting it.\r\n\r\nIn regions using CJK languages, communities share practical challenges such as localization, typography, and translation workflows, creating strong opportunities for regional collaboration.\r\n\r\nThis presentation shares practical experiences from LibreOffice Asia Conference, Wikimedia ESEAP activities, COSCUP, and GNOME Asia Summit. Using examples primarily from LibreOffice and Wikimedia communities, the talk covers what worked and what did not in practice, including co-hosting with other communities, social events, and post-event online communication.\r\n\r\nThe talk also discusses common challenges such as language barriers, limited volunteer capacity, and sustaining collaboration after events.\r\n\r\nThis session offers practical observations on what makes cross-community events effective, and why sustaining connections afterward remains a challenge across Asian open source communities.",{"title":1481,"describe":1482,"type":30},[],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FKLUB3Q",{"id":1487,"room":1488,"start":122,"end":216,"language":96,"track":1489,"speakers":1491,"zh":1506,"en":1509,"tags":1512,"uri":1513},"KMATB3",{"en":39,"zh-hant":10},{"id":384,"name":1490},{"en":386,"zh-hant":10},[1492,1499],{"id":1493,"avatar":1494,"zh":1495,"en":1498},"VR7WAQ","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FVR7WAQ_MbTfjFL.webp",{"name":1496,"bio":1497},"fffuuuming","Web3 Software Engineer Intern @OneSavie Lab\r\nMaster @NTU CSIE",{"name":1496,"bio":1497},{"id":1500,"avatar":1501,"zh":1502,"en":1505},"MWAHKP","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FMWAHKP_mnfPuKW.webp",{"name":1503,"bio":1504},"Aaron Yeh","Web3 Software Engineer Intern @OneSavie Lab\r\nGDG on Campus, Community Lead @University of Taipei \r\nCo-Founder @Corvo AI\r\n\r\n平常就上上學和上上班的大學生，閒暇之餘也創創業、玩玩社群、搞搞區塊鏈。",{"name":1503,"bio":1504},{"title":1507,"describe":1508,"type":30},"當 AI 學會當白帽駭客 – 從鏈上漏洞掃描到資金救援的全自動 Pipeline！","智慧合約的資安事件除了程式碼漏洞本身的問題，更是如何持續監控、快速判斷並即時應對的防禦作業問題。鏈上的威脅會持續出現，留給團隊判斷與應對的時間也和短暫；且「找到一個漏洞」不等於證明它真的可被利用，也不等於能阻止損失發生。\r\n\r\n這場演講將介紹一個一個能自主執行防禦流程的 Agent，目標是打通從漏洞發現到實際行動的完整流程。除了審計原始碼或產生單次的概念驗證（PoC），Agent 更會持續掃描鏈上合約、透過設定好的風險訊號與篩選規則，優先挑出最值得檢查的目標，生成並驗證可利用的 PoC，最終再透過模擬估算真實經濟影響，並在獲得授權時執行資金救援以降低實際損失。\r\n\r\n我們會介紹整體系統架構，並聚焦在這個流程中最困難的一段：「如何從一個潛在漏洞，推進到可實際利用的 PoC，接著再把 PoC 轉換成真正可以部署的救援合約。」我們也會透過 1–2 個可能造成資金損失的具體漏洞案例（e.g., 重入攻擊），討論在 Harness Engineering 的實作上， Agent 該如何推理漏洞的可利用性、取得所需的鏈上狀態、透過模擬驗證攻擊是否成立，以及如何將 PoC 改寫成可執行的救援邏輯。\r\n\r\n我們期待達到的目標，是讓智慧合約安全從週期性的審計，走向持續、端到端的防禦營運；優先處理真正可被利用、具有實質經濟影響，並且能被安全採取行動的風險。",{"title":1510,"describe":1511,"type":30},"When AI Learns to Become a White Hat Hacker: A Fully Automated Pipeline from On-Chain Vulnerability Scanning to Fund Rescue","Smart contract incidents aren’t just a code problem—they’re an operations problem: threats emerge on-chain continuously, time-to-triage is short, and “finding a bug” is not the same as proving exploitability or preventing loss. \r\n\r\nThis talk presents an autonomous security agent designed to close the loop from discovery to action. Instead of only auditing source code or generating standalone proofs-of-concept, it continuously scans on-chain contracts, prioritizes targets using discovery heuristics, generates and validates exploit PoCs, estimates real economic impact via simulation, and performs rescue to reduce loss when authorized.\r\n\r\nWe’ll cover the system architecture, and focus on the most difficult part of the loop: how to move from a suspected vulnerability to an exploitable PoC, and then how to turn that PoC into an actually deployable attack or rescue contract. Through 1–2 concrete examples of vulnerabilities that could lead to fund loss (e.g., Reentrancy), we’ll discuss how the agent reasons about exploitability, gathers the required on-chain context, validates the exploit through simulation, and adapts the PoC into executable rescue logic.\r\n\r\nThe goal is simple: move smart contract security from periodic audits to continuous, end-to-end defense operations that prioritize what is actually exploitable, what matters economically, and what can be acted on safely.",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FKMATB3",{"id":1515,"room":1516,"start":454,"end":731,"language":96,"track":1517,"speakers":1519,"zh":1527,"en":1530,"tags":1531,"uri":1532},"KTZGJD",{"en":237,"zh-hant":10},{"id":241,"name":1518},{"en":243,"zh-hant":243},[1520],{"id":1521,"avatar":1522,"zh":1523,"en":1526},"QN8JDW","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002Fname_DCBBW6Q.png",{"name":1524,"bio":1525},"Ken","Ruby 新手請大家多多指教",{"name":1524,"bio":1525},{"title":1528,"describe":1529,"type":30},"GVL 對 Rails 延遲的實戰觀測 — 用 Vernier 與 gvl-tracing 看懂你的 Puma","你開了 5 個 Puma thread, 真的同時跑了 5 個嗎? MRI 的 GVL 讓答案幾乎永遠是「不」。\r\n這場演講用 Vernier 與 gvl-tracing,把 Rails on Puma 那條看不見的 GVL 等待線直接拍出來給你看",{"title":1528,"describe":1529,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FKTZGJD",{"id":1534,"room":1535,"start":1112,"end":454,"language":70,"track":1536,"speakers":1538,"zh":1553,"en":1556,"tags":1557,"uri":1558},"KWJXTJ",{"en":350,"zh-hant":10},{"id":352,"name":1537},{"en":354,"zh-hant":355},[1539,1546],{"id":1540,"avatar":1541,"zh":1542,"en":1545},"3PT7AJ","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002F3PT7AJ_zRdwl6S.webp",{"name":1543,"bio":1544},"Yongjae Chung","Yongjae is a Master's student at New York University. He is a contributor of gittuf, an incubating project at Open Source Security Foundation.",{"name":1543,"bio":1544},{"id":1547,"avatar":1548,"zh":1549,"en":1552},"XUEYNN","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FXUEYNN_WJ4fq3j.webp",{"name":1550,"bio":1551},"Justin Cappos","I'm a professor at NYU and the creator of five Linux Foundation projects (TUF, in-toto, SBOMit, gittuf, and Uptane).  I also created security architectures used by git and most Linux package managers.",{"name":1550,"bio":1551},{"title":1554,"describe":1555,"type":30},"SBOMs Aren't Enough. Secure Your Software Supply Chain End-To-End","You probably heard that SBOMs are helpful, but did you know that an SBOM only addresses a fraction of what can go wrong in your software supply chain? The SLSA (Supply Chain Levels for Software Artifacts) specification identifies 9 distinct threat areas, spanning from source code, all the way to package distribution. Most development teams address one or two of these and call it a day, leaving gaps that real-world attacks like SolarWinds and Log4J have already exploited. We understand that it is difficult to cover all aspects when it comes to the software supply chain.\r\n\r\nHow about we make this much easier? In this talk, we will present an overview of the modern software supply chain threat model, and show how you can provide integrity throughout the whole process of your software development life cycle. We will introduce an easy-to-setup, end-to-end open source stack, built from frameworks and tools within the CNCF\u002FOpenSSF ecosystem.\r\n\r\nWe will cover:\r\n- gittuf - A security layer for Git repositories\r\n- in-toto - Cryptographically verifiable attestations that capture what actually happens during your build process\r\n- SBOMit - Generate verifiable and accurate SBOMs that embed in-toto attestations\r\n- TUF - A framework for securing package updates\r\n\r\nThis talk aims to demystify software supply chain security beyond SBOMs, and provide a more holistic view. Our live demo will show a pragmatic way to get started, aiming to lower the entry barrier for open source maintainers and adopters alike.",{"title":1554,"describe":1555,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FKWJXTJ",{"id":1560,"room":1561,"start":499,"end":500,"language":70,"track":1562,"speakers":1564,"zh":1572,"en":1575,"tags":1576,"uri":1577},"LBHUAC",{"en":1007,"zh-hant":10},{"id":901,"name":1563},{"en":903,"zh-hant":904},[1565],{"id":1566,"avatar":1567,"zh":1568,"en":1571},"KE3NB8","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FKE3NB8_PfeRvJf.webp",{"name":1569,"bio":1570},"Chi","Chi has spent the last decade developing software for startups focused on data platforms. She played a pivotal role in developing FerretDB, an open-source alternative to MongoDB built atop PostgreSQL. She currently works on the OpenEverest project, where she focuses on automating database provisioning and management on Kubernetes. She is an active contributor to the open-source community, collaborating with peers on GitHub under the handle @chilagrow.",{"name":1569,"bio":1570},{"title":1573,"describe":1574,"type":30},"Running databases on Kubernetes can be easy - meet OpenEverest!","Running databases on Kubernetes is challenging. Choosing Kubernetes to run your workloads has a lot of advantages: it gives you flexibility, and there are plenty of different solutions on the market that can be a part of your perfect technology stack. The real challenge is to build a solid, reliable foundation that will be possible to maintain without causing a headache for your platform team. With databases, it’s even more important to use proven in battle technology, as no one wants to see their data unavailable and their application down. \r\n\r\nTo reduce complexity for database workloads, operators were invented. You can handle the day two operations much more easily than setting up everything by yourself; still, there are different solutions available. To address this challenge, another idea has been born: OpenEverest is a fully open source solution recently donated to CNCF. By running literally three commands, you can set up a tool that allows you to manage different databases and database-related technologies from a unified GUI. You don’t have to worry additionally about backups, monitoring, etc., everything can be controlled from there.\r\n\r\nDuring the presentation, you’ll learn how to use OpenEverest, how to reduce your daily operations, setup your own DBaaS platform, everything available as a 100% open source solution. Additionally, you can learn from case studies how different companies solve their problems with OpenEverest.",{"title":1573,"describe":1574,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FLBHUAC",{"id":1579,"room":1580,"start":476,"end":477,"language":96,"track":1581,"speakers":1583,"zh":1591,"en":1594,"tags":1597,"uri":1598},"LCVJQ8",{"en":922,"zh-hant":10},{"id":924,"name":1582},{"en":926,"zh-hant":927},[1584],{"id":1585,"avatar":1586,"zh":1587,"en":1590},"CVNGF9","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FCVNGF9_hBMqPuy.webp",{"name":1588,"bio":1589},"張仲威","CTO & Co-founder of Canner.",{"name":1588,"bio":1589},{"title":1592,"describe":1593,"type":30},"從 GitHub Stars 到企業客戶：WrenAI 的開源與商業化","WrenAI 是一個開源的 Generative BI 工具，讓使用者用自然語言查詢資料。它從一個想法成長到 GitHub 15k star 的專案，並進入歐美大型企業的正式部署。這場分享會拆解：我們為什麼選擇開源、社群如何成長、以及公司在何時、依據什麼邏輯推出商業版本。",{"title":1595,"describe":1596,"type":30},"From GitHub Stars to Enterprise Customers: Open Source and Commercialization at WrenAI","WrenAI is an open-source Generative BI tool for querying data in natural language. It grew from an idea into a 15k-star GitHub project, now running in production at major enterprises across the US and Europe. This talk breaks down why we chose to open-source, how the community grew, and when and on what logic the company launched a commercial version.",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FLCVJQ8",{"id":1600,"room":1601,"start":382,"end":453,"language":328,"track":1602,"speakers":1604,"zh":1612,"en":1615,"tags":1616,"uri":1617},"LDQXS9",{"en":192,"zh-hant":10},{"id":196,"name":1603},{"en":198,"zh-hant":198},[1605],{"id":1606,"avatar":1607,"zh":1608,"en":1611},"URGXP8","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FURGXP8_j6cdWqC.webp",{"name":1609,"bio":1610},"鄭聖文","Sheng-Wen (Colin) Cheng is a system software engineer with the Platform Security Controller (PSC) team at NVIDIA, where he develops secure bootloader software to ensure system integrity during the boot process. Beyond his role at NVIDIA, he builds low-level software at the intersection of robotics, control engineering, and real-time systems.\r\n\r\nHe is also an active open-source contributor and has presented his work at leading open-source events, including Open Source Summit organized by the Linux Foundation.",{"name":1609,"bio":1610},{"title":1613,"describe":1614,"type":30},"The Illusion of Hardware: How Modern Virtual Machines Work","This talk provides an overall introduction to virtualization technology. The speaker begins with the basic concepts and a brief history of virtualization, introducing virtual machines and hypervisors. We then introduce memory virtualization and address translation, including virtual memory, paging, and the roles of the MMU and TLB.\r\n\r\nNext, we show how virtual machine executes guest system's instructions through direct execution, enabled by hardware-assisted virtualization such as KVM and Virtualization.framework. We contrast this with emulation approaches, including interpretation and JIT compilation.\r\n\r\nFinally, we discuss I\u002FO virtualization, focusing on VirtIO. We show how emulators and hypervisors such as QEMU and ARCN efficiently virtualize devices, including disks, input devices, and graphics, through VirtIO.\r\n\r\nBy the end of this session, the audience will have a clear mental model of how modern VMs and containers work under the hood.",{"title":1613,"describe":1614,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FLDQXS9",{"id":1619,"room":1620,"start":123,"end":709,"language":96,"track":1621,"speakers":1623,"zh":1631,"en":1634,"tags":1637,"uri":1638},"LG9FZH",{"en":1253,"zh-hant":10},{"id":901,"name":1622},{"en":903,"zh-hant":904},[1624],{"id":1625,"avatar":1626,"zh":1627,"en":1630},"PSHHPD","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FScreenshot_2023-07-13_224607_5fH9lhA.png",{"name":1628,"bio":1629},"tobychui","在台灣半導體產業工作的香港人，同時也是開源軟硬體開發者。曾開發多個中小型開源專案，包括 Zoraxy（Reverse Proxy，5.1k☆）與 ArozOS（NAS Web Desktop OS，2.8k☆）等。\r\n\r\nA Hong Konger working in Taiwan’s semiconductor industry, and an open-source software and hardware developer. Author of multiple mid-sized open-source projects, including Zoraxy (a reverse proxy for homelabs, 5.1k☆) and ArozOS (a web desktop OS for NAS systems, 2.8k☆).\r\n\r\nGithub：https:\u002F\u002Fgithub.com\u002Ftobychui\r\nHomepage: https:\u002F\u002Fimuslab.com\u002F",{"name":1628,"bio":1629},{"title":1632,"describe":1633,"type":30},"明明只是想遠端控制我的主機，結果一不小心就做出一套開源 USB \u002F IP KVM 的這件事","在企業環境中，BMC 搭配 IPMI \u002F Redfish 可做到非常完整的 light-out management 功能，但這些功能在 Homelab 使用的家用平台中幾乎不存在。近年市面上雖然有出現很多廉價的 IP KVM ，但大部分都只能控制單一設備，或是需要透過特殊而且設置複雜的 KVM switch 來遠端切換設備。對於有多台設備的 Homelabber 而言無論是軟體設置或是硬體取得上都十分困難。\r\n\r\n本分享介紹開源專案 DezKVM，從 USB KVM（DezKVM-Go）專案出發，延伸至 Web-based IP KVM，並透過 USB Tree 與 firmware 設計打造可擴展的多設備架構。同時涵蓋 MCU 韌體、Linux v4l2 \u002F alsa 以及 Golang 串流實作。\r\n\r\n這不只是 KVM，而是一個可 scale 的遠端控制架構。\r\nhttps:\u002F\u002Fgithub.com\u002Ftobychui\u002FDezKVM\r\nhttps:\u002F\u002Fgithub.com\u002Ftobychui\u002FDezKVM-Go",{"title":1635,"describe":1636,"type":30},"I Only Wanted Remote Access to My Server — Somehow It Turned Into an Open-Source USB\u002FIP KVM Project","In enterprise environments, BMC (Baseboard Management Controller) combined with protocols like IPMI and Redfish enables full-featured lights-out management. However, these capabilities are almost nonexistent on consumer-grade hardware commonly used in homelab setups.\r\n\r\nIn recent years, many low-cost IP KVM solutions have emerged. While they work well for controlling a single device, they often fail to scale. Managing multiple machines either requires complex setups or specialized KVM switches, which are often expensive and difficult to configure. For homelab users with multiple nodes, both the software setup and hardware availability become significant challenges.\r\n\r\nThis talk introduces the open-source project DezKVM, starting from a USB KVM implementation (DezKVM-Go) and evolving into a web-based IP KVM solution. By leveraging USB topology (tree traversal) and firmware-level design, DezKVM enables a scalable architecture for managing multiple devices.\r\n\r\nThe talk will also cover:\r\n- MCU firmware development\r\n- Video and audio capture using Linux V4L2 and ALSA\r\n- Real-time streaming implementation using Golang\r\n\r\nLearn more: \r\nhttps:\u002F\u002Fgithub.com\u002Ftobychui\u002FDezKVM-Go\r\nhttps:\u002F\u002Fgithub.com\u002Ftobychui\u002FDezKVM",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FLG9FZH",{"id":1640,"room":1641,"start":122,"end":216,"language":96,"track":1642,"speakers":1644,"zh":1652,"en":1655,"tags":1657,"uri":1658},"LXY7QK",{"en":1007,"zh-hant":10},{"id":901,"name":1643},{"en":903,"zh-hant":904},[1645],{"id":1646,"avatar":1647,"zh":1648,"en":1651},"FDXVNQ","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FFDXVNQ_or1sn0O.webp",{"name":1649,"bio":1650},"Nick Huang(黃耀德)","喜歡研究linux kernel 以及compiler 的愛好者",{"name":1649,"bio":1650},{"title":1653,"describe":1654,"type":30},"研究 Linux Kernel ISA 架構移植的心路歷程","1. 為什麼會研究 Linux Kernel ISA 架構移植\r\n2. 介紹ISA 架構移植大概的有哪些元件組成\r\n3. 架構移植的流程",{"title":1656,"describe":1654,"type":30},"The research process of adding new archetypes to Linux kernel.",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FLXY7QK",{"id":1660,"room":1661,"start":604,"end":94,"language":96,"track":1662,"speakers":1664,"zh":1675,"en":1678,"tags":1681,"uri":1682},"LZABSB",{"en":684,"zh-hant":10},{"id":686,"name":1663},{"en":688,"zh-hant":689},[1665,1668],{"id":131,"avatar":132,"zh":1666,"en":1667},{"name":134,"bio":135},{"name":134,"bio":135},{"id":1669,"avatar":1670,"zh":1671,"en":1674},"CZHHLS","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FCZHHLS_IZ8HOJO.webp",{"name":1672,"bio":1673},"Denny Huang","SITCON 學生計算機年會共同發起人，活躍於眾多開源社群；在正規教育中特立獨行，妄想透過改善教育讓台灣更進步，崇尚開源精神，熱愛使用者界面以外的多項技術。",{"name":1672,"bio":1673},{"title":1676,"describe":1677,"type":59},"一起來聊聊數位發展部的《公部門開源軟體應用參考手冊》吧！","2026\u002F4\u002F2 由數位發展部、勤業眾信聯合會計師事務所、中華民國軟體自由協會和財團法人開放文化基金會共同打造的《公部門開源軟體應用參考手冊》正式在數位發展部官網上線，紙本也分送到全國公立圖書館，標誌了臺灣公部門擁抱開放原始碼的重要里程碑。\r\n\r\n這份手冊是怎麼誕生的？\r\n預計達到什麼效益？\r\n社群又如何看待他的出現？\r\n\r\n因此我們特別邀請了製作手冊的數位發展部與勤業眾信聯合會計師事務所代表，以及讓手冊在社群廣為人知的夥伴 Denny Huang，一起透過主持人提問、講者交互提問和觀眾提問的方式，談談對於《公部門開源軟體應用參考手冊》的看法與期許。",{"title":1679,"describe":1680,"type":59},"Let's talk about the Public Sector Open Source Software Playbook!","On April 2, 2026, the Public Sector Open Source Software Playbook, jointly created by the Ministry of Digital Affairs, Deloitte Taiwan, the Free Software Foundation Taiwan, and the Open Culture Foundation, officially went live on the Ministry of Digital Affairs' website. Print copies were also distributed to public libraries nationwide, marking an important milestone in Taiwan's public sector embracing open source.\r\n\r\nHow did this playbook come to be?\r\nWhat outcomes does it aim to achieve?\r\nAnd how has the community responded to its arrival?\r\n\r\nTo explore these questions, we have specially invited representatives from the Ministry of Digital Affairs and Deloitte Taiwan — the teams behind the playbook — as well as Denny Huang, who helped bring the playbook to widespread attention within the community. Together, through a format of moderator questions, cross-speaker discussion, and audience Q&A, we will share perspectives and hopes for the Public Sector Open Source Software Playbook.",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FLZABSB",{"id":1684,"room":1685,"start":382,"end":453,"language":96,"track":1686,"speakers":1688,"zh":1696,"en":1699,"tags":1702,"uri":1703},"MCWCLS",{"en":67,"zh-hant":10},{"id":479,"name":1687},{"en":481,"zh-hant":481},[1689],{"id":1690,"avatar":1691,"zh":1692,"en":1695},"UFGYUX","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FUFGYUX_nReSj4p.webp",{"name":1693,"bio":1694},"Wig Cheng","一個開源軟體開發愛好者, 主攻Linux\u002FAndroid BSP開發, 常在AOSP, Linux Kernel upstream做微薄貢獻. 也是台灣Open-EP, 日本Kakip社群的維護者",{"name":1693,"bio":1694},{"title":1697,"describe":1698,"type":30},"open-ep: 從硬體設計到 Mainline Kernel, 一個 e-paper 社群的完整開源實踐","市面上的 e-paper 模組大多依賴廠商提供的 SPI userspace 函式庫，且大多都是MCU的應用，無法與 Linux 標準顯示堆疊整合。更關鍵的是，這些模組的硬體設計和機構細節往往不透明，開發者只能被動接受。\r\nopen-ep 是一個「從硬體到內核全面開源」的社群。我們的 PixPaper 系列產品：\r\n\r\n1. 硬體與機構自主設計\r\n- PCB layout、connector 選型、機構設計全部由社群完成，不依賴 ODM\r\n2. DRM driver upstream\r\n- 已進入 mainline Linux kernel，並獲得 LWN.net 報導\r\n- 因為upstream審核很耗時, 我們也有在vendor kernel實做DRM driver\r\n3. 多 OS 生態支援\r\n- 針對ARM Yocto BSP、Android AOSP、Ubuntu OS, freeRTOS 等不同系統特性擴充應用場景\r\n4. 資安問題納入考慮\r\n- 加入BSP的應用，我們還是會考慮到SBOM的建置與CVE漏洞修補，以符合不同國家(如歐盟CRA)的軟體資安標準\r\n\r\n軟體這不只是「把 driver 推進 kernel」，而是完整展示一個小規模社群如何：\r\n\r\n從電路設計開始就考慮 mainline 相容性\r\n與 kernel maintainer 溝通，讓 patch 順利進入 upstream\r\n平衡「賣產品維持營運」與「推動開源標準化」的雙重目標\r\n透過 COSCUP、UBUCON ASIA, OSPN Japan等技術社群建立信任\r\n\r\n我們相信，硬體開源不只是「公開 schematic」，而是從設計階段就融入開源生態的思維。open-ep 想證明，即使是資源有限、需要自願貢獻者幫忙的社群，也能對 mainline Linux 和產業標準做出實質貢獻。",{"title":1700,"describe":1701,"type":30},"open-ep: From Hardware Design to Mainline Kernel — A Complete Open Source Practice for E-Paper Community","Most e-paper modules on the market rely on vendor-provided SPI userspace libraries and are primarily designed for MCU applications, making them incompatible with the standard Linux display stack. More critically, the hardware design and mechanical details of these modules are often opaque, leaving developers with no choice but to accept them as-is.\r\nopen-ep is a community practicing \"full-stack open source from hardware to kernel.\" Our PixPaper product line features:\r\n\r\n1. Self-designed Hardware and Mechanical Structure\r\n- PCB layout, connector selection, and mechanical design are all completed by the community without relying on ODMs\r\n\r\n2. DRM Driver Upstream\r\n- Merged into mainline Linux kernel and covered by LWN.net\r\n- Since upstream review is time-consuming, we also implement DRM drivers in vendor kernels\r\n\r\n3. Multi-OS Ecosystem Support\r\n- Extended application scenarios across ARM Yocto BSP, Android AOSP, Ubuntu OS, FreeRTOS, and other systems\r\n\r\n4. Security Considerations\r\n- For BSP applications, we address SBOM generation and CVE vulnerability patching to comply with software security standards in different regions (such as the EU CRA)\r\n\r\nThis goes beyond \"pushing drivers into the kernel\" — it demonstrates how a small-scale community can:\r\n\r\nConsider mainline compatibility from the circuit design phase\r\nCommunicate with kernel maintainers to successfully upstream patches\r\nBalance \"selling products for sustainability\" with \"promoting open source standardization\"\r\nBuild trust through technical communities like COSCUP, UBUCON ASIA and OSPN Japan.\r\n\r\nWe believe hardware open source is not just about \"publishing schematics,\" but integrating open source ecosystem thinking from the design stage. open-ep aims to prove that even resource-limited communities relying on volunteer contributions can make substantial contributions to mainline Linux and industry standards.",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FMCWCLS",{"id":1705,"room":1706,"start":216,"end":217,"language":96,"track":1707,"speakers":1709,"zh":1717,"en":1720,"tags":1723,"uri":1724},"MMRBXE",{"en":167,"zh-hant":10},{"id":169,"name":1708},{"en":171,"zh-hant":172},[1710],{"id":1711,"avatar":1712,"zh":1713,"en":1716},"VWXHUR","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FVWXHUR_aiYkdjx.webp",{"name":1714,"bio":1715},"galaxy4552","Hi, I'm galaxy4552, an independent developer exploring alternative approaches to language models and semantic retrieval.\r\nThis talk simply shares some discoveries from my experiments with the open-source project PipeOwl.\r\n\r\n**_The Universe Speaks in Numbers._**",{"name":1714,"bio":1715},{"title":1718,"describe":1719,"type":30},"另一種 Encoding 方法：將語意檢索從O(n²)變成O(n)","目前多數語意搜尋與 embedding 系統，通常依賴 Transformer 或向量相似度計算，其複雜度常接近 O(n²)。在大型語料或即時搜尋場景中，這會帶來顯著的計算成本。\r\n\r\n本分享將介紹一種不同的 encoding 思路，透過結構化語意表示與詞彙場（semantic field）的概念，將查詢與詞彙評分的複雜度降低至 O(n)。\r\n\r\n演講內容包含：\r\n\r\n- 傳統 embedding \u002F Transformer 檢索的計算瓶頸\r\n- 一種替代的語意 encoding 架構\r\n- PipeOwl 專案的設計與實作\r\n- 多語言模型（中文 \u002F 日文）的實驗結果\r\n- 與 BM25、Embedding、FAISS 的速度比較\r\n\r\n本方法與模型已開源，並提供實際 benchmark 與實作程式碼，期待與開源社群討論另一種語意檢索的可能方向。\r\n\r\n相關專案\r\n- https:\u002F\u002Fhuggingface.co\u002FWangKaiLin\u002FPipeOwl-1.6-tw\r\n- https:\u002F\u002Fhuggingface.co\u002FWangKaiLin\u002FPipeOwl-1.5-jp\r\n- https:\u002F\u002Fhuggingface.co\u002FWangKaiLin\u002FPipeOwl-1.4-multilingual",{"title":1721,"describe":1722,"type":30},"An Alternative Encoding Method: Reducing Semantic Retrieval from O(n²) to O(n)","Modern semantic retrieval systems are typically built on Transformer embeddings and vector similarity search, which often result in near O(n²) computational behavior.\r\n\r\nThis talk introduces PipeOwl, an alternative encoding method that treats semantic representation as a structured scoring field over vocabulary, enabling O(n) query evaluation.\r\n\r\nI will discuss the design motivations, implementation details, and benchmark results comparing PipeOwl with BM25, traditional embeddings, and FAISS-based retrieval systems.\r\n\r\nThe project is fully open source and includes multilingual models for Chinese and Japanese.",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FMMRBXE",{"id":1726,"room":1727,"start":709,"end":68,"language":96,"track":1728,"speakers":1730,"zh":1738,"en":1741,"tags":1744,"uri":1745},"MMVBVY",{"en":93,"zh-hant":10},{"id":98,"name":1729},{"en":100,"zh-hant":101},[1731],{"id":1732,"avatar":1733,"zh":1734,"en":1737},"S3HPPG","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FS3HPPG_kqTNwyz.jpeg",{"name":1735,"bio":1736},"范聖佑","JetBrains 技術傳教士，負責推廣 Kotlin、JetBrains IDE 及相關技術。平時醉心於技術研究，專注於 Web、後端技術及能提升生產力的技巧，致力於將複雜的技術名詞用通俗易懂的方式讓新手也能吸收。工作之餘也是多年技術社群志工，積極參與開源社群運作。",{"name":1735,"bio":1736},{"title":1739,"describe":1740,"type":30},"老闆~打包一份技能外帶！讓你的 Kotlin 開發經驗帶著走","實際導入 AI Agent 後，用起來常會發現：跑程式碼風格檢查、靜態分析、執行自動化流程等這些重複出現的開發任務，每次還是要靠自己下 Prompt。即便大型語言模型越來越能理解開發者的意圖，團隊的開發知識卻難以穩定、可攜帶且低成本地傳遞給 Agent。\r\n\r\n這場分享將從開發者與 AI 協作的演進談起，說明為什麼 Agent Skill 會成為下一個值得 Kotlin 工程師投入的 Agent 協作工具鏈。Skill 可以把重複的流程、工具使用方式與團隊慣例，打包成可重複使用的能力，讓 Agent 在需要時才載入相關知識，減少 Token 浪費，也降低每次都重新教 AI 的成本。\r\n\r\n接著，將以 Kotlin 開發者熟悉的情境，介紹 Skill 的基本架構設計、不同 Agent 與開發環境中的套用範圍和管理差異，並討論 Kotlin 開發生態中，有哪些開發知識及情境適合整理成 Skill。最後，會分享如何透過開源工具，讓 Skill 不再綁死在單一提供商、機器或專案，最大化 Skill 帶來的效益。",{"title":1742,"describe":1743,"type":30},"Skills To Go! Bringing Your Kotlin Knowledge Across AI Agents","After adopting an AI agent, you often run into the same friction: recurring development tasks such as running code style checks, static analysis, and automated workflows still require you to write prompts manually. Even as large language models get better at understanding developers' intent, team development knowledge remains difficult to pass to an agent in a stable, portable, and low-cost way.\r\n\r\nThis talk traces the evolution of developer and AI collaboration and explains why Agent Skills are becoming the next tool worth investing in for Kotlin engineers. A Skill packages repeatable workflows, tool usage, and team conventions into a reusable unit, so the agent loads the right knowledge only when needed, reducing token waste and lowering the cost of teaching AI the same things over and over.\r\n\r\nWe will then walk through scenarios familiar to Kotlin developers to introduce the basic architecture of Skills, how scope and management differ across agents and development environments, and what kinds of knowledge and use cases in the Kotlin ecosystem are worth turning into Skills. Finally, we will look at how open-source tooling can keep Skills from being locked to a single vendor, machine, or project, maximizing the value that Skills can bring.",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FMMVBVY",{"id":1747,"room":1748,"start":604,"end":476,"language":96,"track":1749,"speakers":1751,"zh":1759,"en":1762,"tags":1765,"uri":1766},"MN7GQZ",{"en":1154,"zh-hant":10},{"id":330,"name":1750},{"en":332,"zh-hant":332},[1752],{"id":1753,"avatar":1754,"zh":1755,"en":1758},"LNDZBE","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FLNDZBE_qAIyI6h.webp",{"name":1756,"bio":1757},"林博仁","曾經（誤打誤撞）達成百萬安裝數成就的 Snap 發布者。 --- Snap publisher that once (unintentionally) achieved ~1M worldwide active installations.",{"name":1756,"bio":1757},{"title":1760,"describe":1761,"type":59},"Snap 軟體打包入門","- Snap 軟體包簡介\r\n- 適合被打包為 Snap 的軟體\r\n- 不適合被打包為 Snap 的軟體\r\n- Snap 權限限縮模式簡介與它們各自的眉角\r\n- Snap 打包工具簡介\r\n- Snapcraft 專案檔簡介\r\n- Snapcraft 打包生命週期\r\n- Snap 軟體包名稱註冊流程\r\n- Snap 軟體包別名申請流程\r\n- Snap 軟體包維護\r\n\r\nMore info: https:\u002F\u002Fevents.canonical.com\u002Fevent\u002F146\u002Fcontributions\u002F932\u002F",{"title":1763,"describe":1764,"type":59},"Snap packaging 101","* Introduction of snap packages\r\n* Software that is suitable to be packaged as snaps\r\n* Software that is NOT suitable to be packaged as snaps\r\n* Introduction of Snap confinement modes and their respective nuances\r\n* Introduction of snap packaging utilities\r\n* Snapcraft project file introduction\r\n* Snapcraft packaging lifecycle\r\n* Snap package name registration process\r\n* Snap package alias application process\r\n* Snap package maintenance\r\n\r\nMore info: https:\u002F\u002Fevents.canonical.com\u002Fevent\u002F146\u002Fcontributions\u002F932\u002F",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FMN7GQZ",{"id":1768,"room":1769,"start":1770,"end":1175,"language":96,"track":1771,"speakers":1773,"zh":1781,"en":1784,"tags":1787,"uri":1788},"MQX3V9",{"en":407,"zh-hant":10},"2026-08-08T14:45:00+08:00",{"id":410,"name":1772},{"en":412,"zh-hant":412},[1774],{"id":1775,"avatar":1776,"zh":1777,"en":1780},"FTTCJC","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FFTTCJC_w47ertO.webp",{"name":1778,"bio":1779},"Gui 桂","大家好！我是 Gui 桂，目前是一名網頁開發者。\r\n\r\n從高職就讀資處科開始到後來進入職場，深知這個領域需要不斷追求卓越、終身成長，持續往技能更加全面的開發者邁進。\r\n\r\n同時不斷參與社群，參與研討會志工與讀書會，也即將擔任網頁課程老師，希望能將過去所學透過社群分享給更多人。\r\n\r\n目前在公司參與前後端開發，後端以 Node.js + MongoDB 為主軸，這次透過 COSCUP 分享過去在開發會員系統時，整合 Google 第三方登入的真實案例與相關考量，希望能對大家有幫助。",{"name":1778,"bio":1779},{"title":1782,"describe":1783,"type":30},"同個 Email、兩種登入方式 — Node.js + Google 登入整合的四個帳號接管路徑","對任何同時支援「Email + 密碼註冊」跟「Google 登入」或其他第三方登入的系統來說，會員身分可能的狀態組合會比單一登入方式來得複雜，也容易留下一些漏洞。\r\n\r\n本場以一個會員系統(Node.js + Express + TypeScript + MongoDB)為例，拆解四個典型的攻擊手法與對應修補：\r\n\r\n1. 擋下 Google 自動接管已有密碼的帳號\r\n2. Google 登入會員後續該如何安全補上密碼\r\n3. 刪帳號前先解除綁定，避免 orphan binding\r\n4. Pre-account Hijacking：當帳號還沒建立就被卡位的攻擊路徑\r\n\r\n希望能對近期需要整合任何第三方登入的朋友們有一些幫助。",{"title":1785,"describe":1786,"type":30},"One Email, Two Logins — Four Account Takeover Paths When Integrating Google Sign-In with Node.js","For any system that supports both \"email + password\" and \"Google Sign-In\" (or any other third-party login), the possible states of a user identity get noticeably more complex than with a single login method — and that complexity often leaves exploitable holes behind.\r\n\r\nUsing a member system (Node.js + Express + TypeScript + MongoDB) as the case study, this talk walks through four classic account takeover paths and the corresponding fixes:\r\n\r\n1. Blocking Google from silently taking over an account that already has a password\r\n2. How a Google-only user can safely add a password later\r\n3. Unbinding before account deletion to avoid orphan bindings\r\n4. Pre-account hijacking — the attack path that strikes before the account is even created\r\n\r\nHope this serves as a useful reference for anyone currently integrating any kind of third-party login.",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FMQX3V9",{"id":1790,"room":1791,"start":1365,"end":453,"language":96,"track":1792,"speakers":1794,"zh":1798,"en":1801,"tags":1804,"uri":1805},"MWCUSR",{"en":9,"zh-hant":10},{"id":15,"name":1793},{"en":17,"zh-hant":18},[1795],{"id":459,"avatar":460,"zh":1796,"en":1797},{"name":462,"bio":463},{"name":462,"bio":463},{"title":1799,"describe":1800,"type":59},"一起來翻譯！translatewiki × MediaWiki翻譯功能工作坊","本議程解說各項MediaWiki概念，實際操作translatewiki及其他使用MediaWiki翻譯功能並發布譯文，期望引領各位進入各MediaWiki專案\u002F網站的翻譯世界。\r\n本議程涉及實際操作，建議自行攜帶筆記型電腦或平板電腦。",{"title":1802,"describe":1803,"type":59},"Let's Translate! translatewiki × MediaWiki Translate Feature Workshop","This session explains the concepts of MediaWiki, practices with translatewiki and other sites using the MediaWiki Translate feature and publish the translations, hope to introduce the world of internationalization\u002Flocalization of MediaWiki sites.\r\nThis session has practices, we suggest you to bring your own laptops or tablets.",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FMWCUSR",{"id":1807,"room":1808,"start":12,"end":41,"language":96,"track":1809,"speakers":1811,"zh":1815,"en":1818,"tags":1819,"uri":1820},"MWF3N7",{"en":263,"zh-hant":10},{"id":44,"name":1810},{"en":46,"zh-hant":47},[1812],{"id":268,"avatar":269,"zh":1813,"en":1814},{"name":47,"bio":271},{"name":47,"bio":271},{"title":1816,"describe":1817,"type":59},"「匿名支付」主題議程：匿名網路社群與 Blockchain and Distributed Ledger 合辦議程主題","跨社群合作：匿名網路社群 × ETHTaipei\r\n\r\n今年我們與 ETHTaipei（台北以太坊社群）展開議程合作。兩個社群在匿名支付主題上各有不同切入角度，為讓投稿者的稿件能到最合適的聽眾面前。\r\n\r\n這個時段請前往 TR-511 教室參與議程！",{"title":1816,"describe":1817,"type":59},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FMWF3N7",{"id":1822,"room":1823,"start":1824,"end":1825,"language":328,"track":1826,"speakers":1828,"zh":1836,"en":1839,"tags":1840,"uri":1841},"N87PMR",{"en":192,"zh-hant":10},"2026-08-08T12:50:00+08:00","2026-08-08T13:20:00+08:00",{"id":196,"name":1827},{"en":198,"zh-hant":198},[1829],{"id":1830,"avatar":1831,"zh":1832,"en":1835},"KWUVTZ","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FKWUVTZ_rkOcaSk.webp",{"name":1833,"bio":1834},"Yung-Tse Cheng (Mes)","Yung-Tse Cheng is an undergraduate student at National Taiwan Normal University (NTNU), Taiwan. His interests include system software, programming language theory, game engine development, and GPU\u002Fgraphics systems.",{"name":1833,"bio":1834},{"title":1837,"describe":1838,"type":30},"How a VM Draws Pixels: Mesa, Gallium, and virtio-gpu 3D","Drawing a spinning cube on a physical machine sounds ordinary: an application calls a graphics API, the GPU renders the scene, and the result appears on the screen. But when the same program runs inside a virtual machine, the question becomes more interesting. The guest behaves as if it owns a GPU, while the real rendering resources usually live on the host. So where is the cube actually drawn?\r\n\r\nThis talk follows one 3D rendering path inside a Linux guest, starting from Mesa. Mesa receives rendering requests from userspace applications, organizes graphics state and resources through Gallium, and passes the work into VirGL, a driver path designed for virtual GPUs. From there, the work crosses the boundary between Linux DRM and virtio-gpu 3D, becoming contexts, resources, command streams, and synchronization events that a virtual GPU can understand. Finally, the host renderer receives those commands and turns them into pixels.\r\n\r\nUsing a small but complete rendering demo as the thread, we will connect the journey from guest userspace, through the guest kernel and virtual GPU device, to the host renderer. The audience will see that 3D graphics in a VM is not just framebuffer copying. It is a collaboration across Mesa, DRM, virtio-gpu 3D, and the host GPU. Once these boundaries are opened up, “drawing pixels” inside a virtual machine becomes much more interesting than the image that finally appears on screen.",{"title":1837,"describe":1838,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FN87PMR",{"id":1843,"room":1844,"start":499,"end":500,"language":70,"track":1845,"speakers":1847,"zh":1855,"en":1858,"tags":1859,"uri":1860},"N89AEH",{"en":282,"zh-hant":10},{"id":286,"name":1846},{"en":288,"zh-hant":288},[1848],{"id":1849,"avatar":1850,"zh":1851,"en":1854},"FCHQXC","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FFCHQXC_qVXjSbk.JPG",{"name":1852,"bio":1853},"Koji Annoura","Koji Annoura is a Knowledge Graph architect and open-source community leader based in Japan.\r\n\r\nHe has been active in open-source communities for over 20 years, organizing developer meetups and user groups such as the Neo4j Users Group Tokyo and the Apache Hop User Group Japan.\r\n\r\nHis work focuses on graph technologies, knowledge design, SQL\u002FPGQ, GQL, and Context Graphs for connecting knowledge, evidence, decisions, and time.\r\n\r\nHe regularly speaks at international conferences and shares reproducible, hands-on approaches for building knowledge graphs, traceable knowledge systems, GraphRAG systems, and open knowledge workflows.",{"name":1852,"bio":1853},{"title":1856,"describe":1857,"type":30},"From JOINs to Graph Thinking: Practical SQL\u002FPGQ in PostgreSQL","## Description\r\n\r\nModern data is highly connected — such as supply chains, networks, and relationships.\r\n\r\nPostgreSQL is strong for relational queries, but when we need to follow multi-step connections, queries quickly become complex. JOINs grow, and the SQL becomes harder to read, maintain, and extend.\r\n\r\nSQL\u002FPGQ (SQL:2023) brings graph query capabilities into SQL. It allows us to describe connections and paths more directly, without leaving PostgreSQL.\r\n\r\nSQL\u002FPGQ is currently under development in PostgreSQL, and early implementations are already available for experimentation.\r\n\r\nIn this talk, I show how queries evolve from simple joins to multi-step paths. We compare traditional SQL and SQL\u002FPGQ, focusing on readability and how we think about the problem.\r\n\r\nUsing a small trade network example, I demonstrate how SQL\u002FPGQ can simplify queries and help us better understand connected data.\r\n\r\n## Outline\r\n\r\n- Problem: complex JOIN queries  \r\n- SQL\u002FPGQ basics (SQL:2023)  \r\n- SQL vs SQL\u002FPGQ examples  \r\n- Graph thinking in PostgreSQL  \r\n- Current status and how to try it  \r\n\r\n## Key Takeaways\r\n\r\n- JOINs become complex for multi-step queries  \r\n- SQL\u002FPGQ simplifies relationship queries  \r\n- Graph thinking improves understanding of data  \r\n- SQL becomes more powerful, not replaced",{"title":1856,"describe":1857,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FN89AEH",{"id":1862,"room":1863,"start":709,"end":68,"language":96,"track":1864,"speakers":1866,"zh":1874,"en":1877,"tags":1880,"uri":1881},"NENUAS",{"en":237,"zh-hant":10},{"id":241,"name":1865},{"en":243,"zh-hant":243},[1867],{"id":1868,"avatar":1869,"zh":1870,"en":1873},"H9YTTJ","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002Ficon_BM0SbM8.jpg",{"name":1871,"bio":1872},"Cindy","大家好，我是 Cindy，目前在 DEVCORE 當開發工程，7 年以上 Ruby on Rails 開發經驗，喜歡 Ruby，想跟大家分享學到的東西。",{"name":1871,"bio":1872},{"title":1875,"describe":1876,"type":30},"從 2026 年 3 月 Rails 安全更新看三個常見的設計陷阱","2026 年 3 月 23 日，Rails 一次釋出 7.2.3.1、8.0.4.1、8.1.2.1 三個版本，修補了多個 CVE，涵蓋 XSS、ReDoS、記憶體耗盡、路徑穿越、glob 注入等漏洞，是近年規模最大的一次。\r\n本場演講不會逐一拆解 CVE，而是退一步看：這些漏洞之間是否藏著共同的設計陷阱？我會挑出最具代表性的幾個，歸納成三類：信任邊界錯置（內部狀態與外部輸入混用）、檔案系統 API 接受未過濾輸入（path traversal 與 glob injection）、以及「修一次不夠」的不完整修補。\r\n每個陷阱會搭配實際漏洞程式碼與修補方式解說，並提供可立即套用的檢查清單。聽眾不需要資安背景，只要寫過 Rails 應用就能理解。",{"title":1878,"describe":1879,"type":30},"Three Common Design Pitfalls Behind Rails' March 2026 Security Update","On March 23, 2026, Rails released versions 7.2.3.1, 8.0.4.1, and 8.1.2.1, patching multiple CVEs spanning XSS, ReDoS, memory exhaustion, path traversal, and glob injection — its largest security update in recent years.\r\nRather than walking through each CVE, this talk takes a step back: are there shared design pitfalls underneath? I'll pick the most representative cases and group them into three patterns: misplaced trust boundaries (mixing internal state with external input), filesystem APIs accepting unfiltered input (path traversal and glob injection), and \"one fix isn't enough\" — incomplete patches that get bypassed.\r\nEach pitfall comes with vulnerable code, how it was fixed, and a checklist you can apply directly to your project. No security background required.",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FNENUAS",{"id":1883,"room":1884,"start":604,"end":40,"language":1469,"track":1885,"speakers":1887,"zh":1895,"en":1898,"tags":1901,"uri":1902},"NGSU8G",{"en":1063,"zh-hant":10},{"id":1065,"name":1886},{"en":1067,"zh-hant":1068},[1888],{"id":1889,"avatar":1890,"zh":1891,"en":1894},"VRBHKF","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002Fprofile_for_oscon_Kj4txNz.jpg",{"name":1892,"bio":1893},"Masafumi Ohta","Masafumi is now leading the Raspberry Pi community in Japan and helping businesses with Raspberry Pi, especially in Asia. He has also been moderating the OSPN-Japan track at COSCUP since 2019.\r\nMasafumi is not trying new business with Sake\u002FWine breweries to introduce Japanese 'brewing' culture with IT to overseas visitors to Japan; he wants them to discover new things during their visit.",{"name":1892,"bio":1893},{"title":1896,"describe":1897,"type":59},"[BoF] How Do You Grow Your Own OSS Project? — Sustaining, Scaling, and Building Contributors 如何養大你自己的開源專案？ — 一場關於維持、擴展與培養貢獻者","Today, anyone with a small idea can start their own open-source project. The rise of AI-assisted development and the growing population of engineers have made it easier than ever. Yet GitHub now holds a near-endless number of repositories, and the real challenge begins after the first commit:  \r\n**how do you keep your project from being buried, attract users, grow a contributor base, and keep improving?**\r\nFor young engineers especially, this is a daily worry. Running a project demands far more than writing code — visibility, documentation, issue and PR handling, contributor onboarding, and sustaining your own motivation. These problems are rarely solved by technical skill alone, and the hard-won knowledge of those who came before is seldom shared openly.\r\nThis Birds-of-a-Feather session brings together engineers who already maintain an OSS project, those who hope to start one, and those who want to contribute. Rather than a one-way talk, it is a place to bring our struggles and tactics and discuss them together. As a starting point, the facilitator will draw on personal experience leading a project under OpenSolaris and helping several other OSS communities, where many of these problems were faced firsthand. From there, we will dig into the difficulty of maintaining, expanding, and improving a project, and the particular challenge of attracting contributors across borders. To seed the discussion, we will explore questions such as how you won your first users and attention among countless projects, what works for growing contributors and making onboarding smooth, how you deal with critical or demanding participants and turn them into contributors, how you find collaborators globally across language, time-zone, and cultural barriers, and how you avoid burnout and keep a project alive over the long term. Young engineers are the primary audience, but experienced maintainers are warmly welcomed — their stories are exactly what makes this session valuable. No prior knowledge is required.\r\nAttendees will hear candid, real-world stories of how project owners hit walls and worked through them. You will see that your own struggles are not unique, gain concrete tactics you can try the next day, and leave with connections to peers who understand the work of running a project. The facilitator joins as a participant too, eager to learn from everyone's experiences. Whether you maintain a thriving project, a quiet one, or only a plan in your head, you will come away with a clearer sense of how to handle the non-code side of open source — and you will not feel alone in it.",{"title":1899,"describe":1900,"type":59},"如何養大你自己的開源專案？ — 一場關於維持、擴展與培養貢獻者的 BoF","如今，只要有一個小小的點子，任何人都能開始自己的開源專案。AI 輔助開發的興起與工程師人口的增加，讓這件事比以往任何時候都更容易。然而 GitHub 上如今有著幾乎無窮無盡的儲存庫，真正的挑戰其實是在第一次 commit 之後才開始：\r\n你該如何讓自己的專案不被淹沒、吸引使用者、培養出一群貢獻者，並持續改進？\r\n對年輕工程師而言，這尤其是每天都在煩惱的課題。經營一個專案所需要的，遠遠不只是寫程式——還有能見度、文件、Issue 與 PR 的處理、貢獻者的引導上手，以及維持自己的動力。這些問題很少能單靠技術能力解決，而前人辛苦累積下來的經驗，又往往很少被公開地分享出來。\r\n\r\n這場 Birds-of-a-Feather（BoF）聚會，將把已經在維護開源專案的工程師、希望開始一個專案的人，以及想要參與貢獻的人聚在一起。這不是一場單向的演講，而是一個讓大家把各自的難題與心得帶來、一起討論的場合。作為起點，主持人會分享自己在 OpenSolaris 底下帶領一個專案、以及協助其他多個開源社群的親身經驗，這些問題很多都是當時第一線親自面對過的。接著，我們會深入探討維持、擴展與改進一個專案的困難，以及跨越國界吸引貢獻者這個特別的挑戰。為了引導討論，我們會一起探討幾個問題，例如：在數不清的專案之中，你是如何贏得最初的使用者與關注？哪些做法有助於增加貢獻者、並讓新人順利上手？面對提出批評或要求很多的參與者時，你如何應對，又如何把他們轉化為貢獻者？在語言、時區與文化的隔閡之下，你如何在全球尋找協作者？以及，你如何避免倦怠、讓專案能夠長久維持下去？本場次以年輕工程師為主要對象，但也非常歡迎經驗豐富的維護者參加——正是他們的故事，讓這場討論格外有價值。不需要任何先備知識。\r\n\r\n參加者將能聽到專案擁有者如何撞牆、又如何走過難關的真實故事。你會發現自己的掙扎並非獨一無二，帶走一些明天就能嘗試的具體做法，也結識一群同樣懂得經營專案甘苦的夥伴。主持人同樣是以一名參與者的身分加入，期待從每個人的經驗中學習。無論你正在維護一個生氣勃勃的專案、一個安安靜靜的專案，或者腦中還只有一個構想，你都會帶著更清楚的方向離開，知道該如何面對開源中「程式碼以外」的那一面——而且不再覺得孤單。",[],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FNGSU8G",{"id":1904,"room":1905,"start":94,"end":95,"language":96,"track":1906,"speakers":1908,"zh":1916,"en":1919,"tags":1920,"uri":1921},"NHBKPE",{"en":282,"zh-hant":10},{"id":286,"name":1907},{"en":288,"zh-hant":288},[1909],{"id":1910,"avatar":1911,"zh":1912,"en":1915},"DJ3WYV","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FDJ3WYV_mzI9AGc.jpg",{"name":1913,"bio":1914},"後端里長伯","暱稱： 後端里長伯\r\n經歷：\r\n    - sudo rm -rf \u002Fold_mindset\r\n    - Bug 搏鬥倖存者。\r\n排程策略： 12:00 PM 部署（精準落實中午準則）。\r\n備註：系統早晚會出事，所以我只在中午做事。目前的目標是讓 AI 早晚替我做事。",{"name":1913,"bio":1914},{"title":1917,"describe":1918,"type":30},"阿卡西架構：利用 PostgreSQL 打造知識原子管理與閉環演化","「阿卡西紀錄」被視為宇宙中存儲一切知識與事件的虛擬圖書館。在 AI Agent 的開發浪潮中，我們發現單純的資料檢索已不足以應對複雜的工程邏輯。我以此為靈感開發了「阿卡西架構 (Akashic Architecture)」——其目標不再只是「儲存」紀錄，而是將知識轉化為具備執行能力、驗證能力與演化能力的「知識原子」。\r\n\r\n隨著大型語言模型（LLM）的應用邁向複雜系統開發（如前後端整合），傳統的 Agent 擴充模式正遇到擴展性的瓶頸。開發者往往發現，針對特定領域開發的執行單元，在面對高度變動且具備強依賴關係的技術細節時，難以兼顧精準度與系統靈活性。\r\n\r\n本議程將介紹 「阿卡西架構（Akashic Architecture）」，這是一種以 PostgreSQL 為核心樞紐的高階設計模式。我們不再將執行邏輯硬編碼於任務單元中，而是將焦點轉向「動態環境感知」與「執行單元演化」。",{"title":1917,"describe":1918,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FNHBKPE",{"id":1923,"room":1924,"start":499,"end":500,"language":42,"track":1925,"speakers":1927,"zh":1935,"en":1938,"tags":1941,"uri":1942},"NMWE8R",{"en":899,"zh-hant":10},{"id":901,"name":1926},{"en":903,"zh-hant":904},[1928],{"id":1929,"avatar":1930,"zh":1931,"en":1934},"GB9LTU","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FGB9LTU_uNbSfxJ.webp",{"name":1932,"bio":1933},"Steven Yang","Full-Stack Engineer at NTUCOOL, the digital learning platform of National Taiwan University, where I build and maintain production systems serving 30,000+ learners and instructors. My work spans scalable web applications, including video systems, collaborative PDF annotation, course management tools, and AI-assisted features.\r\n\r\nI specialize in TypeScript-based full-stack development, working with modern frontend frameworks and backend services to deliver reliable, maintainable, and user-focused products. I enjoy collaborating with SREs, designers, and product owners to turn complex requirements into stable, real-world systems.\r\n\r\nAlongside my engineering work, I also have experience designing hands-on AI programs and applied prototypes in the education field. This background strengthens my ability to translate complex technical concepts into practical, usable solutions.\r\n\r\nI’m motivated by products that create real impact at scale, especially in education-related domains.",{"name":1932,"bio":1933},{"title":1936,"describe":1937,"type":30},"教學管理平台擴充工具的開放標準：談 LTI 1.3 的機制、應用與實踐","**[ 互通性（Interoperability）是什麼？為什麼 LMS 需要互通性標準？]**\r\n\r\n* 我們將從現代數位學習環境談起：LMS 已不再只是單一平台，而是需要整合影片、測驗、AI、教材、作業與外部服務的學習入口。\r\n\r\n* 探討互通性標準如何降低整合成本，讓不同學習工具能以一致方式與 LMS 協作。\r\n\r\n\r\n**[ 簡介 1EdTech 的互通性生態系 ]**\r\n\r\n* 介紹 IMS Global 到 1EdTech 的演進背景，以及其在教育科技領域的角色。\r\n\r\n* 簡介主要互通性標準：LTI（Learning Tools Interoperability）、QTI （Question and Test Interoperability）、One Roster 說明這些標準如何支援身份驗證、跨平台整合。\r\n\r\n\r\n**[ 探究 LTI 1.3 的核心運作概念 ]**\r\n\r\n* LTI 1.1 到 LTI 1.3 的演進：新機制、新功能與安全性改善。\r\n\r\n* 教學平台與擴充工具如何溝通：LTI 1.3 Launch Flow （OIDC Login、OAuth 2.0、JWT \u002F id_token）\r\n\r\n**[ Demo：以 NTUCOOL \u002F Canvas LMS 為例串接 LTI 1.3 學習工具 ]**\r\n\r\n* 實際展示：如何在 Canvas 建立與設定 LTI 1.3 Tool？他有什麼明顯的好處？\r\n\r\n* 經驗分享：談 NTUCOOL 如何依照現有的技術堆疊實踐 LTI 1.3",{"title":1939,"describe":1940,"type":30},"Open Standards for LMS: Mechanisms, Applications, and Implementations of LTI 1.3","## [ What Is Interoperability, and Why Do LMS Platforms Need Interoperability Standards? ]\r\n\r\n* We begin with the modern digital learning environment: LMS platforms are no longer standalone systems, but gateways that integrate videos, assessments, AI tools, learning materials, assignments, and external services.\r\n\r\n* We explore how interoperability standards reduce integration costs and enable different learning tools to collaborate with LMS platforms in a consistent and standardized manner.\r\n\r\n## [ Introduction to the 1EdTech Interoperability Ecosystem ]\r\n\r\n* An introduction to the evolution from IMS Global to 1EdTech and its role in the educational technology ecosystem.\r\n\r\n* Overview of major interoperability standards, including LTI (Learning Tools Interoperability), QTI (Question and Test Interoperability), and OneRoster, and how these standards support authentication and cross-platform integration.\r\n\r\n## [ Exploring the Core Concepts Behind LTI 1.3 ]\r\n\r\n* The evolution from LTI 1.1 to LTI 1.3: new mechanisms, enhanced capabilities, and improved security.\r\n\r\n* How LMS platforms and external tools communicate through the LTI 1.3 Launch Flow, including OIDC Login, OAuth 2.0, and JWT \u002F id_token validation.\r\n\r\n## [ Demo: Integrating an LTI 1.3 Learning Tool with NTUCOOL \u002F Canvas LMS ]\r\n\r\n* Live demonstration: how to create and configure an LTI 1.3 tool in Canvas, and the key benefits it provides.\r\n\r\n* Experience sharing: how NTUCOOL implemented LTI 1.3 based on its existing tech stack.",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FNMWE8R",{"id":1944,"room":1945,"start":1946,"end":1947,"language":96,"track":1948,"speakers":1950,"zh":1958,"en":1961,"tags":1962,"uri":1963},"NNHVVW",{"en":1253,"zh-hant":10},"2026-08-08T10:55:00+08:00","2026-08-08T11:25:00+08:00",{"id":901,"name":1949},{"en":903,"zh-hant":904},[1951],{"id":1952,"avatar":1953,"zh":1954,"en":1957},"XLSPGW","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FXLSPGW_esdGoMB.webp",{"name":1955,"bio":1956},"Jalen","一個沒事愛亂嘗試，愛各種瘋瘋癲癲點子的網路工程師",{"name":1955,"bio":1956},{"title":1959,"describe":1960,"type":30},"LanParty千人打遊戲，用開源Router打造全亞洲最大網咖","WirForce 是台灣每年舉辦的大型 LanParty，2024 年擁有 1200 個 BYOC 座位，是亞洲規模最大的同類活動。要讓超過千名玩家同時在線、低延遲打遊戲，網路基礎設施必須在有限的佈建時間內達到「盡可能不讓玩家抱怨」的穩定性。\r\n2024 年，我首次加入現有網路團隊，尋找開源的方案來處理Router。本議程會誠實覆盤當時發生了什麼、走過多少地雷，以及這些教訓如何成為 2025 年的改進目標。\r\n2025 年我們改用 VyOS 搭配 VPP（Vector Packet Processing），提供 800 個 BYOC 座位的網路使用。分享在這種大型看似單一簡單的架構中網路的困難與碰到甚麼問題，以及用了哪些方法來減輕網路負擔。\r\n如果你正在評估 VyOS \u002F VPP 用於高密度臨時網路，或只是想知道「千人 LanParty 的網路會出什麼事」，這場分享會給你一個來自現場的經驗分享。",{"title":1959,"describe":1960,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FNNHVVW",{"id":1965,"room":1966,"start":239,"end":408,"language":96,"track":1967,"speakers":1969,"zh":1977,"en":1980,"tags":1983,"uri":1984},"NSFLQL",{"en":9,"zh-hant":10},{"id":15,"name":1968},{"en":17,"zh-hant":18},[1970],{"id":1971,"avatar":1972,"zh":1973,"en":1976},"ZXUDGZ","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FZXUDGZ_W18nhBO.webp",{"name":1974,"bio":1975},"Yi-Jyun Pan","Git 繁體中文 l10n 團隊的 Lead，也是中華民國軟體自由協會 Weblate 實例（Tryneeds）的維護者。",{"name":1974,"bio":1975},{"title":1978,"describe":1979,"type":30},"Git 繁體中文翻譯流程治理實戰","Git 是幾乎所有開發者每天都會碰到的工具，但你可能很少想過：它的繁體中文翻譯是怎麼被維護的？\r\n\r\n這場分享不會把重點放在「某個詞到底該怎麼翻」的語言學辯論。實際上，Git 繁體中文翻譯社群很小，真正困難的往往不是翻譯決策，而是更務實的問題：**要怎麼讓人願意開始翻？要怎麼 review？要怎麼把大家的貢獻整理成符合 Git 上游慣例的提交？**\r\n\r\nGit l10n 有一套自己的運作方式。翻譯不是隨便一個人改完 `.po` 檔、送 PR 就結束；要注意提交到哪個 repo、對應的 commit 格式（Signed-off-by）、squash、寄送到 l10n 流程，以及最後如何進入 Git release。這些規則對熟悉 Git 的人來說都不一定直覺，更不用說只是想幫忙修一句中文的新貢獻者。\r\n\r\n為了降低門檻，我們導入 Weblate，希望讓不熟悉 Git 流程的人也能直接在網頁上參與翻譯。但導入之後才發現，Weblate 產生 commit 的方式，和 git-l10n 對提交記錄的要求並不完全相容。於是，維護者必須在「讓貢獻變簡單」與「讓 upstream 可以接受」之間，設計出一套實際可運作的中間流程。\r\n\r\n本議程會分享 Git 繁體中文翻譯團隊實際踩過的坑：最初是如何從簡體中文基礎建立繁體中文翻譯、如何吸引新貢獻者、如何 review 翻譯品質、如何處理 Weblate 產生的 commit、如何透過人工 squash、Co-Authored-By 與自動化腳本保留貢獻紀錄，最後再把整理好的成果送進 Git 的正式 l10n 流程。\r\n\r\n這場議程也會分享，一個翻譯字串從 Weblate 被修改、被維護者 review、被整理成符合慣例的 commit，到最後進入 Git release，中間到底經過哪些關卡。也會談到，在人力有限的情況下，哪些事情值得自動化，哪些事情仍然需要人工判斷，以及維護者如何在貢獻者體驗與上游規範之間取得平衡。\r\n\r\n如果你維護過開源專案、想建立翻譯流程、正在思考如何降低貢獻門檻，或只是好奇「一句中文翻譯要怎麼正式走進 Git」，這場議程會提供一個非常真實、很小但很完整的案例。",{"title":1981,"describe":1982,"type":30},"Practical Workflow Governance for Git Traditional Chinese Localization","Git is a tool that almost every developer touches every day, but you may not have thought much about how its Traditional Chinese translation is maintained.\r\n\r\nThis talk is not about linguistic debates over “what is the perfect translation for this one term.” In reality, the Git Traditional Chinese localization community is small, and the hardest problems are usually not about translation decisions. They are much more practical: **how do we get people to start contributing? How do we review translations? And how do we turn everyone’s contributions into submissions that follow Git’s upstream localization conventions?**\r\n\r\nGit l10n has its own way of working. Translating Git is not as simple as having someone edit a `.po` file and open a PR. You need to know which repository to submit to, what commit format is expected, how `Signed-off-by` should be handled, when and how to squash commits, how changes are sent through the l10n process, and eventually how they make it into a Git release. These rules are not always intuitive even for people who already use Git, let alone new contributors who just want to fix a single Chinese string.\r\n\r\nTo lower the barrier to entry, we introduced Weblate so that people who are not familiar with the Git contribution workflow could still help translate directly in a web interface. But after adopting it, we found that the way Weblate generates commits does not fully align with git-l10n’s expectations for commit history. As a result, maintainers had to design a practical middle layer between “making contribution easy” and “making the result acceptable upstream.”\r\n\r\nIn this talk, I will share the real-world pitfalls the Git Traditional Chinese localization team has run into: how the translation was initially bootstrapped from the Simplified Chinese version, how we try to attract new contributors, how we review translation quality, how we deal with commits generated by Weblate, how we preserve contributor credit through manual squashing, `Co-authored-by`, and automation scripts, and how we eventually send the cleaned-up result into Git’s official l10n process.\r\n\r\nThe talk will also walk through the full journey of a translated string: from being edited in Weblate, reviewed by maintainers, reorganized into a commit that follows upstream conventions, and finally shipped in a Git release. Along the way, I will discuss what is worth automating when the team has limited capacity, what still requires human judgment, and how maintainers balance contributor experience with upstream requirements.\r\n\r\nIf you maintain an open source project, want to build a translation workflow, are thinking about how to lower the barrier for contributors, or are simply curious about how one Chinese sentence officially makes its way into Git, this talk offers a small but complete real-world case study.",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FNSFLQL",{"id":1986,"room":1987,"start":1278,"end":499,"language":328,"track":1988,"speakers":1990,"zh":1998,"en":2001,"tags":2002,"uri":2003},"PEDPRR",{"en":326,"zh-hant":10},{"id":330,"name":1989},{"en":332,"zh-hant":332},[1991],{"id":1992,"avatar":1993,"zh":1994,"en":1997},"FG7JPJ","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FFG7JPJ_pRzpJKC.webp",{"name":1995,"bio":1996},"Tushar Gupta","Tushar Gupta is a Prof 1 at Canonical, where he works on the Launchpad team, building the infrastructure that powers Ubuntu. Passionate about open-source, he loves to spend time on OSS projects and has worked on p5py (Processing for Python), Flagsmith (feature flag management), and Sarthi (Docker-based preview environments). He is also an alumnus of the MLH Fellowship, Processing Summer Fellowship, and the GitHub Campus Experts program.",{"name":1995,"bio":1996},{"title":1999,"describe":2000,"type":30},"Creating containers like Michelangelo","AI has had an interesting side effect: it has shifted security from something often treated as an afterthought into a proactive, first-class concern in modern software development. With LLMs, scanning and identifying attack vectors is no longer limited to the codebase; it now extends across transitive dependencies, third-party packages, and entire distributions, resulting in a steady stream of newly uncovered critical CVEs.\r\n\r\nContainers sit at the core of modern service deployments. Securing them is critical, and one of the most effective ways to strengthen security is to reduce the attack surface by stripping away everything that isn't essential.\r\n\r\nThe challenge is doing that in a reliable, efficient and repeatable way.\r\n\r\nLuckily, there's a tool built for exactly this: Rocks.\r\n\r\nIt's an OCI-compliant image builder that lets you chisel away unnecessary packages and dependencies, helping you sculpt a lean, purpose-built container image, much like Michelangelo revealing a masterpiece from a block of \"rock\". \r\n\r\n(pun intended)\r\n\r\nMore info: https:\u002F\u002Fevents.canonical.com\u002Fevent\u002F146\u002Fcontributions\u002F944\u002F",{"title":1999,"describe":2000,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FPEDPRR",{"id":2005,"room":2006,"start":238,"end":239,"language":96,"track":2007,"speakers":2009,"zh":2017,"en":2020,"tags":2023,"uri":2024},"PFQEET",{"en":1007,"zh-hant":10},{"id":901,"name":2008},{"en":903,"zh-hant":904},[2010],{"id":2011,"avatar":2012,"zh":2013,"en":2016},"DLBWBK","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FDLBWBK_1cYJgte.png",{"name":2014,"bio":2015},"Richard Lyu","Richard Lyu 是 SUSE Labs 的 EFI 開發者，專門從事 edk2\u002FOVMF 及機密運算的開發與研究，同時也是 UEFI Forum 的成員。他畢業於國立陽明交通大學，擁有深厚的低層系統背景，擅長原始碼分析與追蹤。Richard Lyu 曾受邀至國立清華大學的高等 Unix 程式設計課程擔任講者， 也曾至 FOSDEM 2025 、CYBERSEC 2025 發表研究。他熱衷於參與技術社群，曾擔任 HITCON Enterprise 2024 副總召集人，以及 HITCON Training 2023 總召集人。",{"name":2014,"bio":2015},{"title":2018,"describe":2019,"type":30},"x86-64 下機密運算的演進","隨著雲端運算日益普及，資料隱私與機密性已無法僅依賴軟體層的加密。硬體層級的隔離，以及遠端信任機制，如今已成為不可或缺的要素。本演講將追溯 x86-64 平台上機密運算（Confidential Computing）的演進，並著重探討這些技術如何在雲端環境中被採用。\r\n\r\n我們將從 Intel SGX 談起，說明 enclave 如何建立細粒度的信任邊界，接著我們會轉向當今的主流方法：以虛擬機（VM）為基礎的隔離。透過比較 AMD SEV-SNP 與 Intel TDX 的設計，我們將檢視它們在記憶體頁面保護、遠端證明（remote attestation），以及與虛擬化堆疊整合方面的差異。這使既有應用程式能在幾乎不需修改程式碼的情況下，保有硬體層級的機密性，同時在並非完全可信的雲端基礎設施之上，建立可驗證的執行環境。",{"title":2021,"describe":2022,"type":30},"The Evolution of Confidential Computing on x86-64","As cloud computing becomes increasingly widespread, data privacy and confidentiality can no longer rely solely on software level encryption. Hardware based isolation and remote trust mechanisms have become essential. This talk will trace the evolution of confidential computing on the x86-64 platform, with a focus on how these technologies are adopted in cloud environments.\r\n\r\nWe will start with Intel SGX and explain how enclaves establish fine grained trust boundaries. We will then move to today’s mainstream approach: VM based isolation. By comparing the designs of AMD SEV-SNP and Intel TDX, we will examine their differences in memory page protection, remote attestation, and integration with virtualization stacks. This allows existing applications to gain hardware backed confidentiality with little to no code changes, while establishing a verifiable execution environment on top of cloud infrastructure that is not fully trusted.",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FPFQEET",{"id":2026,"room":2027,"start":1155,"end":2028,"language":328,"track":2029,"speakers":2031,"zh":2039,"en":2042,"tags":2043,"uri":2044},"PKYW3X",{"en":192,"zh-hant":10},"2026-08-08T16:00:00+08:00",{"id":196,"name":2030},{"en":198,"zh-hant":198},[2032],{"id":2033,"avatar":2034,"zh":2035,"en":2038},"LSA8G9","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FLSA8G9_ulm3eGw.webp",{"name":2036,"bio":2037},"Chi-Kuan Chiu","NCKU CSIE 115",{"name":2036,"bio":2037},{"title":2040,"describe":2041,"type":30},"Box64 in Action: Playing x86-64 Steam Games on ARM64 \u002F RISC-V with Dynamic Binary Translation","Running unmodified x86-64 binaries on ARM64 and RISC-V\r\nLinux systems remains important, but general-purpose emulators such as\r\nQEMU often impose too much overhead for demanding workloads, while\r\nhardware-assisted translators such as Rosetta 2 depend on platform-\r\nspecific silicon. Box64 addresses this gap with a wrapping-first architec-\r\nture that routes calls into native host libraries through ABI translation\r\nwrappers and reserves JIT compilation for guest code.\r\nThis paper describes three parts of the design: a native wrapping layer\r\nspanning 278 libraries, including entry-point interception and callback\r\nbridging; a four-pass JIT compiler with flag liveness analysis, deferred\r\nflag computation, and native flag reuse; and system-level support for\r\ndeployment, including Steam container bypass, Wine mixed-bitness co-\r\nordination, and TSO emulation on weakly ordered hosts. Evaluation on\r\nRaspberry Pi 5 (ARM64), HiFive™ Premier P550 and Andes AX45MP \r\n(RISC-V) shows a median of 67% native CPU performance on ARM64 \r\nmicrobenchmarks, 63–94% native throughput for favorable wrapped \r\nlibrary calls, and a median 4× speedup over QEMU, reaching \r\n40×when wrapping removes translation from the hot path.",{"title":2040,"describe":2041,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FPKYW3X",{"id":2046,"room":2047,"start":1112,"end":454,"language":70,"track":2048,"speakers":2050,"zh":2065,"en":2068,"tags":2069,"uri":2070},"PRKPG7",{"en":1007,"zh-hant":10},{"id":901,"name":2049},{"en":903,"zh-hant":904},[2051,2058],{"id":2052,"avatar":2053,"zh":2054,"en":2057},"FWUFZL","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FFWUFZL_t5WLmEb.webp",{"name":2055,"bio":2056},"leenahyun","Hello, I am Nahyun Lee, a student at Ajou University in South Korea. I founded and operated a student community that runs an OpenStack-based campus cloud. I am interested in creating practical infrastructure environments for students and in building sustainable communities with open-source technologies.",{"name":2055,"bio":2056},{"id":2059,"avatar":2060,"zh":2061,"en":2064},"7XW8XF","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002F7XW8XF_RKOctFq.webp",{"name":2062,"bio":2063},"Hwagyun Kim","I am a founder of Aolda, student of Ajou University, South Korea. With the interest of infrastructure, my expertise interest lies in Network Security Architecture with building and securing OpenStack cloud environments.",{"name":2062,"bio":2063},{"title":2066,"describe":2067,"type":30},"Can Students Build Their Own Cloud? What Open Source Gave Us, and What It Didn’t","This talk begins with a simple question: if students are to truly learn cloud infrastructure, what kind of environment should we give them?\r\n\r\nAt Aolda, a student cloud infrastructure group at Ajou University in South Korea, we did not want to stop at handing out server accounts. We wanted students to see how a cloud is put together, try things on their own, and learn by touching the systems behind the service. That goal led us to open source. Commercial solutions were expensive and convenient, but they also hid too much of the structure we wanted students to understand. So we built our environment around OpenStack, MicroCeph, Proxmox VE, and observability tools such as Prometheus, Grafana, and Loki.\r\n\r\nOpen source gave us many of the things we were looking for: flexibility, visibility, and room to learn by doing. At the same time, it exposed us to problems that are discussed less often. Documentation was uneven, learning paths were unclear, and some tools solved one problem while creating another. We learned that open source is not a shortcut around complexity. It changes the kind of work you have to do, and it forces you to think more carefully about what you are really trying to build.\r\n\r\nThat experience shaped not only our infrastructure, but also our student community itself. We had to decide what was realistic for students to learn, what level of complexity was worth keeping, and how to build a culture around shared study, experimentation, and responsibility.\r\n\r\nThis talk is not a tutorial or a polished success story. It is a practical account of why we chose open source, what it helped us build, where it fell short, and how those trade-offs shaped the direction of our group. We hope it will give students and small communities a more honest starting point for building with open source.",{"title":2066,"describe":2067,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FPRKPG7",{"id":2072,"room":2073,"start":95,"end":1155,"language":70,"track":2074,"speakers":2076,"zh":2084,"en":2087,"tags":2088,"uri":2089},"PU8GEB",{"en":899,"zh-hant":10},{"id":901,"name":2075},{"en":903,"zh-hant":904},[2077],{"id":2078,"avatar":2079,"zh":2080,"en":2083},"AZSUCQ","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FAZSUCQ_zGccbLy.webp",{"name":2081,"bio":2082},"Akshat Sharma","Akshat Sharma is a Developer Advocate and GraphQL Ambassador , dedicated to building meaningful connections between technology and community.\r\nIn 2024, he was selected as one of 1,220 contributors from over thousonds of  applicants in Google Summer of Code. In 2025, he returned as a mentor—guiding contributors in Google Summer of Code 2025, Mifos Summer of Code, and Code for GovTech. His research journey has taken him to the University of California and IIT Indore, and he has authored eight research papers in recent advancements and technology. He has presented his work at prestigious conferences including ICACIS 2023, ICNSBT 2024, C3IT 2024, the Open Source Security Summit, Interledger Summit, and the Bharat FinTech Summit. He has also spoken at major global events, delivering talks at the Interledger Summit in Cape Town, GRPC Conf in Bengaluru, and OpenSearchCon in Japan. Additionally, he has attended industry-leading gatherings such as Open SSF, Open Source Summit India, and KubeCon + CloudNativeCon.",{"name":2081,"bio":2082},{"title":2085,"describe":2086,"type":30},"The Internet of Money: Building Open Financial Bridges with Interledger","Despite rapid advances in digital payments, financial systems remain fragmented and exclusionary. This talk explores how the Interledger Protocol (ILP) enables seamless interoperability across diverse payment networks, creating a foundation for open and inclusive finance.\r\n\r\nWe will dive into how Interledger works, real-world use cases, and how open-source contributors can help shape the future of global payments.",{"title":2085,"describe":2086,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FPU8GEB",{"id":2091,"room":2092,"start":499,"end":500,"language":70,"track":2093,"speakers":2095,"zh":2103,"en":2106,"tags":2107,"uri":2108},"PXFF8E",{"en":237,"zh-hant":10},{"id":241,"name":2094},{"en":243,"zh-hant":243},[2096],{"id":2097,"avatar":2098,"zh":2099,"en":2102},"YHA8GQ","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FYHA8GQ_LNzBjJU.webp",{"name":2100,"bio":2101},"Sampo Kuokkanen","Hi Taiwan! My name is Sampo Kuokkanen and I currently live in Osaka, Japan, but I'm originally from Finland. I lived in Taiwan for three years, so I can speak a bit of Chinese too. \r\n\r\nI work as a Ruby on Rails developer and I'm also the head of the Japanese office for Evil Martians, keeping everything going here in Japan. I recently started looking at JRuby after Charles Nutter's talk in RubyKaigi 2026 and I'm planning on talking about that and the larger testing area of Ruby.",{"name":2100,"bio":2101},{"title":2104,"describe":2105,"type":30},"Ruby on Ruby: ruby\u002Fspec, MRI Tests, and JRuby and Living on the Edge","Ruby has multiple implementations. CRuby (the canonical \"MRI\") is the reference, but JRuby (now at Ruby 4.0 on the JVM) and TruffleRuby (at 3.4, with 4.0 underway on GraalVM) are real, production-grade alternatives. They aim to run the same Ruby code, so your gems, your Rails apps, your scripts. But they're written in different languages, on different VMs, with different internal data structures. So how do they agree on what Ruby actually is?",{"title":2104,"describe":2105,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FPXFF8E",{"id":2110,"room":2111,"start":477,"end":2028,"language":96,"track":2112,"speakers":2114,"zh":2122,"en":2125,"tags":2126,"uri":2127},"PXKFX9",{"en":93,"zh-hant":10},{"id":98,"name":2113},{"en":100,"zh-hant":101},[2115],{"id":2116,"avatar":2117,"zh":2118,"en":2121},"KNVAW3","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FKNVAW3_rhuxFtx.webp",{"name":2119,"bio":2120},"Shane Liu(櫛風)","一個莫名其妙學了很多東西的人，現在就讀於清華大學資安所碩一。\r\n退役競程選手，整天寫一些怪東西，歡迎來找我玩～",{"name":2119,"bio":2120},{"title":2123,"describe":2124,"type":59},"打造你自己的 Kotlin DSL：以文字冒險遊戲為例","你有沒有想過，你可以用 Kotlin「發明」一門小語言？\r\n\r\nDSL 簡單來說，就是利用各種語法特性，讓你在設計函式、物件、lambda 時，雖然寫的還是合法的 Kotlin，卻能讓使用者感覺像是在使用一門為特定問題量身打造的語言。\r\n\r\n本 Workshop 將帶你從零開始，用 Kotlin 打造一個文字冒險遊戲的腳本 DSL。從最樸素的 Builder 出發，一步步引入其他語法，最終設計出一個可讀性極高、可直接描述房間、劇情與選項的 Kotlin DSL。\r\n\r\n結束後你不只學到 DSL 怎麼做，更能看懂其他 DSL 背後共同的設計哲學——那些你每天在用的框架，其實都是同一套思路打造的。",{"title":2123,"describe":2124,"type":59},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FPXKFX9",{"id":2129,"room":2130,"start":709,"end":68,"language":96,"track":2131,"speakers":2133,"zh":2141,"en":2144,"tags":2147,"uri":2148},"QCEEEE",{"en":39,"zh-hant":10},{"id":384,"name":2132},{"en":386,"zh-hant":10},[2134],{"id":2135,"avatar":2136,"zh":2137,"en":2140},"NAQNEX","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FNAQNEX_kHlOP29.jpg",{"name":2138,"bio":2139},"Allen Chu","- XueDAO Core Contributor\r\n- 長期參與海內外 Web3 Hackathon 的 Builder \r\n- AI 技術講師，分享 AI 應用實作\r\n- PawX 共同創辦人，提供 Web3 Twitter\u002FX 輿情資料\r\n- 前 Machine Learning Engineer\r\n\r\n目前正在思考人生的意義",{"name":2138,"bio":2139},{"title":2142,"describe":2143,"type":30},"ACP：面向 Onchain AI Agent 經濟的通訊與交易協議","隨著 Agentic Finance 與 Autonomous Agents 的討論度快速升溫，大家開始不再只關注「AI 能不能完成任務」，而是進一步思考：當 AI Agent 可以自己溝通、協商、交易甚至彼此合作時，我們該如何為它們建立一套真正可運作的經濟系統？包含如何在區塊鏈上安全地傳遞訊息、驗證結果、完成付款，以及建立彼此之間的信任與 reputation。\r\nAgent Commerce Protocol（ACP）正是在這樣的背景下，於 2025 年由 Virtuals Protocol 提出的開放協議。ACP 目標是打造一套面向 onchain agent commerce 的基礎設施，讓 buyer、agent 與 seller 能在去中心化環境中進行可驗證的互動與交易。透過智能合約托管、加密驗證、獨立 evaluator 機制，以及多 Agent 協作架構，降低多步驟交易中的信任成本與協作風險。ACP 因此不只是支付協議，更像是一套專為 AI Agent 經濟打造的鏈上通訊與商務協議層。\r\n雖然 ACP 是一年前提出的概念，但其中許多設計至今依然極具前瞻性，甚至能對應目前 AI Agent、MCP、Agent-to-Agent communication 與 onchain coordination 的發展方向。本次 Workshop 將以深入淺出的方式，帶大家理解 ACP 背後的設計理念、Agent 在鏈上的協作模式，以及它如何嘗試建立未來 Agent Economy 的基礎框架。現場也會透過實際案例與 demo，展示多個 AI Agent 如何在鏈上彼此互動、協商與完成任務。",{"title":2145,"describe":2146,"type":30},"ACP: Communication and Commerce Protocol for the Onchain AI Agent Economy","As discussions around Agentic Finance and Autonomous Agents continue to grow rapidly, people are beginning to move beyond asking whether AI can complete tasks, and are instead exploring a deeper question: when AI agents can communicate, negotiate, transact, and collaborate autonomously, how do we build an economic system that truly enables them to operate independently? This includes challenges such as securely exchanging information onchain, verifying outcomes, executing payments, and establishing trust and reputation between agents.\r\n\r\nAgent Commerce Protocol (ACP) emerged in this context as an open protocol introduced by Virtuals Protocol in 2025. ACP aims to create foundational infrastructure for onchain agent commerce, enabling buyers, agents, and sellers to engage in verifiable interactions and transactions within decentralized environments. Through smart contract escrow, cryptographic verification, independent evaluator mechanisms, and multi-agent coordination architectures, ACP reduces trust assumptions and collaboration risks in complex multi-step transactions. As a result, ACP is not merely a payment protocol, but rather a communication and commerce protocol layer specifically designed for the AI agent economy.\r\n\r\nAlthough ACP was proposed over a year ago, many of its ideas remain highly forward-looking today, aligning closely with the current evolution of AI Agents, MCP, Agent-to-Agent communication, and onchain coordination systems. In this workshop, we will explore the core design principles behind ACP in an accessible and practical way, covering how agents collaborate onchain and how ACP attempts to establish the foundational infrastructure for a future Agent Economy. The session will also include live demos and practical examples showcasing how multiple AI agents can interact, negotiate, and complete tasks autonomously onchain.",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FQCEEEE",{"id":2150,"room":2151,"start":95,"end":1155,"language":96,"track":2152,"speakers":2154,"zh":2162,"en":2165,"tags":2168,"uri":2169},"QGUGMK",{"en":39,"zh-hant":10},{"id":44,"name":2153},{"en":46,"zh-hant":47},[2155],{"id":2156,"avatar":2157,"zh":2158,"en":2161},"ZMHFCQ","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FZMHFCQ_nbu4tmQ.jpg",{"name":2159,"bio":2160},"黃豆泥 mashbean","黃豆泥，密碼龐克嚮往者，專注於分散式科技與數位自主權，曾為醫師、數位發展部的約聘公務員，現為 Matters 總經理、哈佛大學政策訪問研究員（Allen Lab Fellow）與國際策略論壇訪問學者（International Strategic Forum 2026 Fellow）。黃彥霖於 2023 年至 2025 年間任職於數位發展部擔任資安制度工程師，推動分散式技術的數位公共建設規劃，具體內容為數位憑證皮夾計畫以及開放網路標準，並參與全球資訊網協會（W3C）。此外，豆泥現為以太坊基金會森友會（Silviculture Society）成員，提供區塊鏈網路治理之諏議，目標為為新興科技尋求良善的網路治理方案。",{"name":2159,"bio":2160},{"title":2163,"describe":2164,"type":30},"我不洗錢，為何要理解匿名支付？從零開始介紹隱私加密金流交易","我不洗錢，為何要理解匿名支付？\r\n\r\n許多人以為加密貨幣是匿名的，事實正好相反。只要知道一個錢包地址，就能在公開鏈上看到對方完整的金流、社會關係與行動軌跡。對於倡議組織、跨境協作的公民團體、需要保護身份的捐款人與受助者來說，這是一個被嚴重低估的風險。\r\n\r\n過去幾年，區塊鏈社群為了讓「隱私」回到鏈上，發展出三套設計風格截然不同的解決方案。最早登場的是 Tornado Cash，技術上是經典的混幣器，但因為缺乏治理機制，最終被美國政府制裁、開發者也遭到起訴。接棒的 Privacy Pool 改採白名單路線，使用者需要取得「匿名良民證」才能進入隱私池，把合規與隱私放進同一個架構下協商。Railgun 則走另一條路，以黑名單搭配冷卻期與集體偵測，在 2025 年成功阻擋了 zkLend 駭客 950 萬美元的洗錢嘗試。最後是剛在以太坊社群冒出的錢包套件 Kohaku，目標是讓「隱私」變成所有錢包的預設值，以集體防禦取代個人選擇。\r\n\r\n這個講座將以最白話的方式走過隱私金流的設計邏輯，並試圖討論對於開源社群與公民團體都很重要的問題，當隱私不只是技術問題，而是治理問題，誰來定義好人與壞人？\r\n\r\n這場講座適合關心數位人權與公民安全的工作者、想了解加密貨幣合規邊界的法遵與會計人員、對隱私技術好奇但不知從何切入的開源開發者，以及任何曾經對「我又沒做壞事，為什麼要在乎隱私」這句話感到不太對勁的人。",{"title":2166,"describe":2167,"type":30},"I Don't Launder Money. Why Should I Care About Anonymous Payments? A Beginner's Guide to Private Crypto Transactions","Many people assume cryptocurrency is anonymous. The reality is the opposite: with just a single wallet address, anyone can trace the holder's complete financial flows, social relationships, and movement patterns across the public chain. For advocacy organizations, civic groups engaged in cross-border collaboration, and donors and recipients who need to protect their identities, this is a severely underestimated risk.\r\n\r\nOver the past few years, the blockchain community has developed three distinct approaches to bringing privacy back on-chain. The first was Tornado Cash, a classic coin mixer that was technically elegant but lacked any governance mechanism — leading to U.S. government sanctions and the prosecution of its developers. Privacy Pool took up the baton with a whitelist-based approach: users must obtain an \"anonymous proof of innocence\" before entering the privacy pool, placing compliance and privacy within the same negotiable framework. Railgun took a different route, combining a blacklist with a cooling-off period and collective detection — a design that successfully blocked the zkLend hacker's attempt to launder $9.5 million in 2025. Finally, Kohaku, a newly emerging wallet toolkit in the Ethereum community, aims to make privacy the default for every wallet, replacing individual choice with collective defense.\r\n\r\nThis talk will walk through the design logic of private financial flows in plain language, and tackle a question that matters deeply to both open source communities and civic groups: when privacy is no longer just a technical problem but a governance problem, who gets to define who is \"good\" and who is \"bad\"?\r\n\r\nThis session is for those working on digital rights and civic safety, compliance and accounting professionals seeking to understand the regulatory boundaries of cryptocurrency, open source developers curious about privacy technology but unsure where to begin, and anyone who has ever felt that something about the phrase \"I have nothing to hide, so why should I care about privacy?\" doesn't quite sit right.",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FQGUGMK",{"id":2171,"room":2172,"start":2173,"end":2174,"language":96,"track":2175,"speakers":2177,"zh":2185,"en":2188,"tags":2191,"uri":2192},"QHRTDA",{"en":407,"zh-hant":10},"2026-08-08T13:35:00+08:00","2026-08-08T14:05:00+08:00",{"id":410,"name":2176},{"en":412,"zh-hant":412},[2178],{"id":2179,"avatar":2180,"zh":2181,"en":2184},"AZCHDZ","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FAZCHDZ_Fqx9kqC.webp",{"name":2182,"bio":2183},"楊騏 Ci","Ci 是全端工程師，軟體開發 10+ 年。iThome 鐵人賽「佛心分享」佳作得主，JSDC 2025、HWDC 2025 講者。\r\n\r\n專注在 AI 驅動開發——怎麼讓 AI 真的照規格寫 code、怎麼把專案知識結構化餵給 AI，而不是每次對話都重講一遍。開源 SDD 工具 Prospec（MIT）作者，每天把自己寫的工具用在自己的專案上。",{"name":2182,"bio":2183},{"title":2186,"describe":2187,"type":30},"我把自己寫的開源 SDD 工具丟到 15 個專案實測：一份 AI 開發 dogfooding 誠實清單","AI 寫 code 很快，但每開一個新對話，它又忘了你的專案長怎樣、用什麼 pattern，於是你又重講一次。工具越來越多，反而越用越累。\r\n\r\n我把這個痛點做成開源工具 Prospec（MIT、TypeScript CLI）：漸進式揭露讓 context 按需載入、Delta Specs 只追蹤變更、AI Knowledge 把專案知識結構化版控，不隨對話視窗消失。\r\n\r\n但這場我不只 demo 工具。過去半年我把它真的用在 15 個專案上，從個人 side project 到公司專案。哪些設計真的省時間、哪些是我過度設計到沒人用；SDD 在既有專案和全新專案完全是兩種體感；我從 Spec-Kit、OpenSpec 抄了什麼、又刻意不抄什麼。\r\n\r\n適合每天跟 AI 寫 code、但受夠重複解釋專案的開發者。你會帶走一套能馬上試的 SDD 流程，和一份「哪些 SDD 招式是噱頭」的誠實清單。",{"title":2189,"describe":2190,"type":30},"Running My Own Open-Source SDD Tool Across 15 Projects: An Honest Dogfooding Report","AI writes code fast, but every new chat it forgets your project's architecture and conventions, so you explain it all over again. More tools, more fatigue.\r\n\r\nI turned that pain into an open-source tool, Prospec (MIT, TypeScript CLI): progressive disclosure loads context on demand, delta specs track only changes, and AI Knowledge keeps project knowledge structured and version-controlled instead of vanishing with the chat window.\r\n\r\nBut this isn't just a tool demo. Over the past six months I've run Prospec across 15 of my own projects, from personal side projects to work codebases: which designs genuinely saved time, which were over-engineering nobody used, how SDD feels completely different on brownfield vs greenfield, and what I borrowed from Spec-Kit and OpenSpec and deliberately didn't.\r\n\r\nFor developers who code with AI daily but are tired of re-explaining their project. You'll leave with an SDD workflow you can try immediately, and an honest list of which SDD tricks are just hype.",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FQHRTDA",{"id":2194,"room":2195,"start":709,"end":68,"language":96,"track":2196,"speakers":2198,"zh":2202,"en":2205,"tags":2206,"uri":2207},"QPDZEC",{"en":263,"zh-hant":10},{"id":44,"name":2197},{"en":46,"zh-hant":47},[2199],{"id":268,"avatar":269,"zh":2200,"en":2201},{"name":47,"bio":271},{"name":47,"bio":271},{"title":2203,"describe":2204,"type":30},"記者如何用開源工具保護消息來源：從第一次接觸到報導刊出後的數位整理","調查報導的風險，最重的一端常落在消息來源身上。記者背後通常還有編輯部與媒體律師，消息來源卻可能是公司內部員工、政府機關承辦或被監控的社運參與者，暴露的代價高得多。這場沿著一個常見流程走一遍，每一步都對應到一套原始碼公開、可被獨立檢視的開源工具：第一次接觸的安全收件管道（Signal、Freedom of the Press Foundation 維運的 SecureDrop、OnionShare 都是開源專案）、如何在不留紀錄的前提下確認對方身分、敏感檔案交換前用 mat2\u002FMetadata Cleaner 去除 metadata 並以 VeraCrypt 或 Cryptpad 加密儲存、訪談紀錄的最小化，以及報導刊出後記者端與消息來源端各自要做的數位整理。這場也說明為什麼這些情境偏好開源工具，加密實作可被公開審查、不必單靠一家商業公司的善意。內容會對照台灣的法規脈絡（通訊保障及監察法、2025 年上路的揭弊者保護法）與在地調查報導圈的實務做法。",{"title":2203,"describe":2204,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FQPDZEC",{"id":2209,"room":2210,"start":1155,"end":2028,"language":328,"track":2211,"speakers":2213,"zh":2221,"en":2224,"tags":2225,"uri":2226},"QPFMND",{"en":326,"zh-hant":10},{"id":330,"name":2212},{"en":332,"zh-hant":332},[2214],{"id":2215,"avatar":2216,"zh":2217,"en":2220},"EVQD9U","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FEVQD9U_ImfU7HU.webp",{"name":2218,"bio":2219},"Rajani Ekunde","Rajani is a Senior Software Engineer (DevOps) with over 4 years of experience in cloud infrastructure, automation,SRE and platform engineering. She is a Google Developer Expert (GDE) for Google Cloud, a Docker Captain, and a Women Techmakers (WTM) Ambassador, actively contributing to the global tech community. Rajani is a frequent speaker at major cloud and DevOps conferences, where she shares practical insights on modern infrastructure and emerging technologies. Passionate about knowledge sharing, Rajani aims to inspire and empower developers to build scalable, reliable systems.",{"name":2218,"bio":2219},{"title":2222,"describe":2223,"type":30},"Invisible Debugging for Docker Workloads on Ubuntu GCP: eBPF in Action","Modern production systems often restrict SSH access, making traditional debugging on Ubuntu instances increasingly difficult. In this talk, I share real-world SRE experiences debugging live issues on Ubuntu virtual machines running on GCP, without logging into them.\r\n\r\nUsing eBPF-based tools on Ubuntu, I demonstrate how to trace system calls, analyze network behavior, and identify performance bottlenecks in real time. Through actual incident scenarios, we’ll uncover how Ubuntu’s kernel capabilities combined with eBPF enable “invisible debugging,” helping engineers resolve critical issues faster without disrupting production systems.\r\n\r\nMore info: https:\u002F\u002Fevents.canonical.com\u002Fevent\u002F146\u002Fcontributions\u002F929\u002F",{"title":2222,"describe":2223,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FQPFMND",{"id":2228,"room":2229,"start":2230,"end":1112,"language":96,"track":2231,"speakers":2233,"zh":2241,"en":2244,"tags":2245,"uri":2246},"QQEZEP",{"en":93,"zh-hant":10},"2026-08-08T13:50:00+08:00",{"id":98,"name":2232},{"en":100,"zh-hant":101},[2234],{"id":2235,"avatar":2236,"zh":2237,"en":2240},"GJA7GD","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FIMG_6998-removebg-preview_Yd3TWKC.png",{"name":2238,"bio":2239},"Andy Lu","Hello 我是 Andy，一名 Android 工程師。在 AI 大爆發時代，也不斷思考著如何將其引入開發中，\r\n期待與各位分享自身與 AI 經驗。",{"name":2238,"bio":2239},{"title":2242,"describe":2243,"type":30},"三招看穿偽裝的 Agent Skill！在 Android 開發中建立 AI 安全防護網","現在大家使用 Claude Code 或 Codex 等 AI 代理來寫 Code、甚至自動幫忙 Build 和上架 Android App。但你知道嗎？高達 **13.4% 的第三方 AI Skill 藏有致命風險**！\r\n\r\n如果你隨便安裝來路不明的 Skill，惡意程式可能就在背景偷偷把你的 Android 簽名金鑰 (`.jks`) 打包傳送給駭客。\r\n\r\n我們將展示如何防範「AI 內鬼」。你將學會 **「防禦三步驟（靜態分析、AI 空跑預測、沙盒誘餌測試）」** 來掃描惡意 Skill；同時，我們也會示範如何化被動為主動，利用 IntelliJ 內建的 MCP 伺服器，讓 AI 代理在寫扣時，自動幫你揪出潛在漏洞並安全地重構程式碼，建立真正的零信任開發環境！",{"title":2242,"describe":2243,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FQQEZEP",{"id":2248,"room":2249,"start":605,"end":476,"language":96,"track":2250,"speakers":2252,"zh":2259,"en":2262,"tags":2263,"uri":2264},"RAFMPA",{"en":192,"zh-hant":10},{"id":196,"name":2251},{"en":198,"zh-hant":198},[2253],{"id":2254,"avatar":3,"zh":2255,"en":2258},"ZAGJW7",{"name":2256,"bio":2257},"Aristo Chen","Aristo Chen is an embedded Linux engineer with a strong interest in low-level systems\r\nprogramming and open-source development.\r\n\r\nusb-proxy is a personal open-source project he built to deepen his understanding of the\r\nLinux USB stack from the ground up. The project combines the Linux kernel's raw-gadget\r\ndriver, libusb, and a Lua scripting engine to create a transparent USB man-in-the-middle\r\nproxy that runs on hardware such as a Raspberry Pi. It has gathered over 250\r\nstars and 40 forks on GitHub.\r\n\r\nBeyond usb-proxy, he actively contributes to upstream open-source projects in the\r\nembedded Linux ecosystem, including U-Boot and the Linux kernel, with a focus on\r\nbootloader enablement and hardware bringup.",{"name":2256,"bio":2257},{"title":2260,"describe":2261,"type":30},"USB MITM: Building a Transparent USB Proxy with raw-gadget and libusb","The USB protocol is everywhere: keyboards, webcams, storage devices, embedded hardware\r\ntest fixtures. Yet most developers treat it as a black box. What if you could sit\r\n*transparently* between a USB host and device, observe every packet, and selectively\r\nmodify, drop, or inject traffic using nothing but open-source software running on a\r\nRaspberry Pi?\r\n\r\nThis talk walks through the design and implementation of\r\n[usb-proxy](https:\u002F\u002Fgithub.com\u002FAristoChen\u002Fusb-proxy), an open-source USB\r\nman-in-the-middle proxy built on the Linux kernel's `raw-gadget` driver and `libusb`.\r\n\r\n## What is usb-proxy?\r\n\r\n`usb-proxy` positions a Linux machine with a USB OTG port between a real USB device and\r\nits host. The host sees a synthetic device reconstructed in real time via `raw-gadget`,\r\nwhile the real device is accessed through `libusb` on the other side. All traffic flows\r\nthrough the proxy, where it can be logged, modified, or dropped using JSON-based rules.\r\n\r\n## The Injection Rule Engine\r\n\r\n`injection.json` defines per-endpoint rules for control, interrupt, and bulk transfers.\r\nEach rule can modify, ignore, or stall a matched packet. Three levels of transform are\r\nsupported and can be combined within a single rule, in order of increasing flexibility:\r\n\r\n- **Pattern replacement**: matches fixed hex byte sequences and substitutes them.\r\n  Ideal for simple, fixed substitutions such as swapping mouse button bytes.\r\n- **Declarative operations**: applies arithmetic transforms in order on byte offsets,\r\n  such as negating a movement axis or scaling cursor speed, without any scripting.\r\n- **Lua scripting**: for logic that cannot be expressed declaratively, a rule can point\r\n  to a Lua script. Each script maintains its own state across packets, enabling\r\n  conditionals, loops, and stateful transforms like dead zones or speed caps.\r\n\r\n## What Will Be Coverd\r\n\r\n- How the Linux USB stack is layered: UDC driver, gadget framework, and userspace\r\n- How `raw-gadget` and `libusb` work together to build a transparent C++ proxy\r\n- The injection rule engine in action with a live demo\r\n\r\n## Talk Outline\r\n\r\n| Section | Time |\r\n|---|---|\r\n| USB protocol and motivation | 5 min |\r\n| raw-gadget + libusb architecture deep-dive | 10 min |\r\n| Injection rule engine: live demo on how to modify and inject USB packets | 10 min |\r\n| QA | 5 min |\r\n\r\n**Source**: https:\u002F\u002Fgithub.com\u002FAristoChen\u002Fusb-proxy (Apache-2.0 license)",{"title":2260,"describe":2261,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FRAFMPA",{"id":2266,"room":2267,"start":216,"end":217,"language":96,"track":2268,"speakers":2270,"zh":2278,"en":2281,"tags":2283,"uri":2284},"RBHMPE",{"en":282,"zh-hant":10},{"id":286,"name":2269},{"en":288,"zh-hant":288},[2271],{"id":2272,"avatar":2273,"zh":2274,"en":2277},"7HLLC8","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002F19250689_10155459347729154_1385127100667476617_o_SOXtwc0.jpg",{"name":2275,"bio":2276},"林宗禧","自PostgreSQL 9 成為重度使用者，初以開發FDW 為樂；於2014-2016 間，約兩年半的時間，聯繫拜訪國內推廣PostgreSQL 的前輩，以及聚集相關技服廠商，自2017 年再次推動台灣PostgreSQL使用者社群(TWPUG) ；目前著重發展 PostgreSQL 為核心的 Data Application Platform，並推廣在各類產業應用。",{"name":2275,"bio":2276},{"title":2279,"describe":2280,"type":30},"Why PostgreSQL is Becoming Huge in AI?","AI 時代 PostgreSQL 的角色正在重新定義",{"title":2282,"describe":2280,"type":30},"Why PostgreSQL is Becoming Huge in AI ?",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FRBHMPE",{"id":2286,"room":2287,"start":2288,"end":122,"language":328,"track":2289,"speakers":2291,"zh":2295,"en":2298,"tags":2299,"uri":2300},"RTQZZR",{"en":1253,"zh-hant":10},"2026-08-08T09:00:00+08:00",{"id":1258,"name":2290},{"en":1260,"zh-hant":1261},[2292],{"id":1264,"avatar":1265,"zh":2293,"en":2294},{"name":1267,"bio":10},{"name":1267,"bio":10},{"title":2296,"describe":2297,"type":30},"COSCUP x UbuCon Asia 2026: Welcome Day 1","COSCUP Opening\r\nUbuCon Asia Introduction\r\nPrime Speaker Introduction",{"title":2296,"describe":2297,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FRTQZZR",{"id":2302,"room":2303,"start":2173,"end":2174,"language":70,"track":2304,"speakers":2306,"zh":2314,"en":2317,"tags":2318,"uri":2319},"RX7U8K",{"en":1253,"zh-hant":10},{"id":901,"name":2305},{"en":903,"zh-hant":904},[2307],{"id":2308,"avatar":2309,"zh":2310,"en":2313},"TNUR7W","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FTNUR7W_wyjrJvs.webp",{"name":2311,"bio":2312},"Vincent Chapuis 沙文森","French Engineer, graduated both from France Engineering school and Taiwanese University, passionate about Open-Source and project making, also enjoys to play and create music. And clearly in love with Taiwan.\r\nOften accompanied by his brother in arms Arthur Amalvy 艾亞瑟, PhD graduated from France and Taiwan, NLP researcher in Academia Sinica. Together, they take on new projects every 2 Mondays and fight for honor in multitudes of video games.",{"name":2311,"bio":2312},{"title":2315,"describe":2316,"type":30},"A Modern Open-Source Stack for Robotics Side Projects","This talk, targeted at beginners, explores the architecture of an open-source toolstack for the making of robotics projects. From the software building tool, to the simulation software to test your robot, we will help you discover Nix, Rust, Embassy, Gitea and its CI, KiCad, FreeCad and Webots and how to use\u002Fconnect them to start working on your next side project!",{"title":2315,"describe":2316,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FRX7U8K",{"id":2321,"room":2322,"start":12,"end":382,"language":1469,"track":2323,"speakers":2325,"zh":2333,"en":2336,"tags":2339,"uri":2340},"RYMNYC",{"en":1063,"zh-hant":10},{"id":1065,"name":2324},{"en":1067,"zh-hant":1068},[2326],{"id":2327,"avatar":2328,"zh":2329,"en":2332},"3VF3BU","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002F338381963_529392195798536_5185507161144438443_n_nuQykxg.jpg",{"name":2330,"bio":2331},"Jun Iio","Dr. Jun Iio was born in Gifu in 1970. He received his master's degree in mathematical engineering and information physics from the University of Tokyo in 1994 and joined Mitsubishi Research Institute, Inc. in the same year. He has been a visiting associate professor at the International Center of Tokyo University of Agriculture and Technology since 2009. In 2013, he became an associate professor at the Department of Socio-informatics, Faculty of Letters, Chuo University; a research fellow at the Institute of Science and Engineering, Institute of Social Science, Chuo University; and, from 2014, a professor at the same department at the same university. He has been a professor at the Faculty of Global Informatics at the same university since 2019. He is also a visiting scholar at King Mongkut's Institute of Technology Ladkrabang, Thailand, in 2025. He is engaged in research on the interaction between humans and information systems. His research keywords include image processing, user interfaces, and the optimization of information systems. He is an HCD-Net-certified human-centered design expert, a professional engineer (information engineering), and holds a Ph.D. in engineering.",{"name":2330,"bio":2331},{"title":2334,"describe":2335,"type":30},"Why We Still Need to Teach App Development in the AI Era and Why FOSS Matters","In the era of generative AI, even novice users can create simple applications through well-crafted prompts. However, although AI tools can generate working code, they do not automatically teach inexperienced developers or students how to read, review, test, license, maintain, or collaborate around that code. Against this backdrop, as software development becomes increasingly automated, an important question arises for both educators and open-source communities: Why is it still important to teach the fundamentals of software development, and what role can FOSS (Free and Open Source Software) play in shaping the next generation of developers?\r\n\r\nThis presentation explores these questions through practical examples from application development projects conducted in the Iio Laboratory at the Faculty of Global Informatics, Chuo University. By engaging students in FOSS-based development projects, the laboratory provides opportunities to learn programming and software engineering fundamentals while experiencing open collaboration, community-driven development, transparency, and shared ownership—values that remain essential even in the age of AI-assisted coding.\r\n\r\nThis talk will be valuable for educators, FOSS community leaders, mentors, and professionals involved in developer training. Participants will gain practical insights into how FOSS functions not only as a development model but also as an educational environment that cultivates developers capable of making meaningful contributions to open communities.",{"title":2337,"describe":2338,"type":30},"在AI時代，我們為何仍需要教授應用程式開發，以及FOSS為何重要？","在生成式人工智慧時代，即使是新手用戶也能透過精心設計的提示創建簡單的應用程式。然而，儘管人工智慧工具可以產生可運行的程式碼，但它們並不會自動教導新手開發者或學生如何閱讀、審查、測試、授權、維護或圍繞這些程式碼進行協作。在這種情況下，隨著軟體開發日益自動化，一個重要的問題擺在了教育者和開源社群面前：為什麼教授軟體開發的基礎知識仍然很重要？自由開源軟體（FOSS）在塑造下一代開發者方面又能發揮怎樣的作用？\r\n\r\n本次演講將透過中央大學國際資訊學部Iio Lab所進行的應用程式開發專案實例來探討這些問題。 Iio Lab為學生提供參與基於FOSS的開發專案的機會，讓他們不僅學習程式設計和軟體工程的基礎知識，還能體驗在人工智慧輔助編碼時代仍然至關重要的價值觀，例如開放協作、社群驅動開發、透明度和共享所有權。\r\n\r\n本次演講將對教育工作者、FOSS社群領導者、導師以及參與開發者培訓的專業人士有所裨益。參與者將獲得關於 FOSS 如何不僅作為一種開發模式，而且作為一個強大的教育環境來培養未來能夠為開放社區做出有意義貢獻的開發者的實際見解。",[],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FRYMNYC",{"id":2342,"room":2343,"start":238,"end":239,"language":96,"track":2344,"speakers":2346,"zh":2354,"en":2357,"tags":2359,"uri":2360},"S8HXVH",{"en":93,"zh-hant":10},{"id":98,"name":2345},{"en":100,"zh-hant":101},[2347],{"id":2348,"avatar":2349,"zh":2350,"en":2353},"QZCYVZ","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FQZCYVZ_uw9V0ms.webp",{"name":2351,"bio":2352},"Kyle Lin","A compiler \u002F programming language enthusiast.",{"name":2351,"bio":2352},{"title":2355,"describe":2356,"type":30},"駕馭 Scala 型別系統：在型別層打造自動機","Scala 的型別系統在 JVM 生態系中獨樹一幟，但也因為其強大的功能讓開發者時常不知道如何運用這個優勢去開發型別上更加安全的程式。\r\n\r\n在這個 session 中，將會簡單介紹 Scala 型別系統，並利用其底層的推導邏輯打造一個型別安全的自動機，初步帶領開發者理解型別系統並開發的擁有進階型別安全的程式。",{"title":2358,"describe":2356,"type":30},"Exploiting Scala Type System: Building Type-Level Automaton",[402],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FS8HXVH",{"id":2362,"room":2363,"start":499,"end":500,"language":96,"track":2364,"speakers":2366,"zh":2374,"en":2377,"tags":2380,"uri":2381},"S8MFLU",{"en":167,"zh-hant":10},{"id":169,"name":2365},{"en":171,"zh-hant":172},[2367],{"id":2368,"avatar":2369,"zh":2370,"en":2373},"TLWDMZ","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FTLWDMZ_BN7DNfj.webp",{"name":2371,"bio":2372},"Max","目前就讀於成功大學資訊工程系",{"name":2371,"bio":2372},{"title":2375,"describe":2376,"type":30},"在 FPGA 上從 RTL 到 Token 走過 LLM 推理全流程","我透過 DE10-Nano FPGA 開發板，沒有 GPU、不呼叫任何框架，從模型量化、推論引擎到 RTL 硬體設計，在 AI 推理的每一層都讓自己做中學，將 BitNet 2B transformer 模型塞在 1GB 記憶體的 DE10-nano FPGA 上完成 token 輸出。\r\n\r\n此演講適合那些想要摸清 LLM 從底層硬體到最終 token 輸出的人。\r\n\r\n這場分享會聊：三元權重 LLM 為何特別適合 FPGA、T-MAC 查表的實作與資源平衡、HPS\u002FFPGA 混合推理引擎的設計取捨、以及在資源極度受限硬體上做 LLM 部署的踩坑經驗。",{"title":2378,"describe":2379,"type":30},"From RTL to Token: Walking Through the Complete LLM Inference Pipeline on an FPGA","Using a DE10-Nano FPGA development board—with no GPU and no frameworks—I built every layer of the AI inference stack myself as a learning-by-doing journey: from model quantization and the inference engine down to RTL hardware design, ultimately getting a BitNet 2B transformer model to generate tokens on the DE10-Nano.\r\nThis talk is for anyone who wants to understand how an LLM works all the way from the underlying hardware to the final token output.\r\nTopics covered: why ternary-weight LLMs are a natural fit for FPGAs, implementing T-MAC lookup tables and balancing hardware resources, design trade-offs in the HPS\u002FFPGA hybrid inference engine, and lessons learned from deploying an LLM on extremely resource-constrained hardware.",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FS8MFLU",{"id":2383,"room":2384,"start":12,"end":382,"language":96,"track":2385,"speakers":2387,"zh":2395,"en":2398,"tags":2400,"uri":2401},"SAVQ9E",{"en":684,"zh-hant":10},{"id":686,"name":2386},{"en":688,"zh-hant":689},[2388],{"id":2389,"avatar":2390,"zh":2391,"en":2394},"GF3AUJ","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002Fchenghao-yang_Z8AqEJo.jpg",{"name":2392,"bio":2393},"梯口 tico88612","CNCF 推廣大使（CNCF Ambassador）、Cloud Native Taiwan Co-organizer、Kubernetes 貢獻者（Kubespray 維護者、Release Team v1.32~v1.37）。",{"name":2392,"bio":2393},{"title":2396,"describe":2397,"type":30},"開源是免費的嗎？從 CNCF 大使視角看人才、企業、政策三方困境","「開源」常被當成「免費」的同義詞 —— 公司想省授權費、政府想省採購成本、工程師想省學習門檻。但維護一套被全世界依賴的開源專案，從來不是免費的。\r\n\r\n本場次以 CNCF 大使與 CNTUG 社群組織者的視角，拆解一個在台灣很少被正面談論的問題：當開源早已是雲端原生時代的產業標準，台灣為什麼始終站在「使用者」而非「主導者」的位置？講者將從人才、企業、政策三個面向切入：\r\n* 人才：刷開源缺乏舞台、薪資沒有溢價，優秀貢獻者最後多半往國外出走；\r\n* 企業：用賣硬體的思維做軟體、靠高度客製化的代工生存，把「開源會被別人學走」當成不開源的理由——卻忘了若 Kubernetes 當年不開源，今天就不會有整個 CNCF 生態讓大家共享；\r\n* 政策：政府傾向「辦活動」而非資助真正在爬坡的貢獻者。對照 Python 基金會為了堅持 DEI 價值、主動撤回美國 NSF 一百五十萬美元補助，缺口最終由企業補上的案例，我們得以重新檢視「政府主導開源」的侷限。\r\n\r\n如果台灣能解開這三方的結構性困境，它能釋放的影響力，或許會超越我們對自己的想像。",{"title":2399,"describe":2397,"type":30},"Is open source really free? A CNCF Ambassador's perspective on the talent, corporate, and policy trilemma",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FSAVQ9E",{"id":2403,"room":2404,"start":122,"end":216,"language":42,"track":2405,"speakers":2407,"zh":2415,"en":2418,"tags":2421,"uri":2422},"SCJ33F",{"en":899,"zh-hant":10},{"id":901,"name":2406},{"en":903,"zh-hant":904},[2408],{"id":2409,"avatar":2410,"zh":2411,"en":2414},"THHM39","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FTHHM39_VwHidrT.webp",{"name":2412,"bio":2413},"石獅 ciosai","演算藝術家，用 opengl 或 webgl 等底層工具創作 demoscene 作品\r\nmeteoriks 2025 新秀獎得主\r\n平常也喜歡畫圖和編曲",{"name":2412,"bio":2413},{"title":2416,"describe":2417,"type":30},"開源即興程式表演：哪裡找資源和社群？","音樂表演背景的酷選視覺，到數學邏輯演出的音樂，你有聽過 \"即興程式\"(livecoding) 嗎？一起來看看有什麼開源的工具可以馬上入坑！\r\n\r\nhttps:\u002F\u002Fgithub.com\u002Ftoplap\u002Fawesome-livecoding\u002Fblob\u002Fmaster\u002FREADME.md\r\n\r\nhttps:\u002F\u002Fvcvrack.com\u002F https:\u002F\u002Ftixl.app\u002F https:\u002F\u002Fcables.gl\u002F https:\u002F\u002Fstrudel.cc\u002F https:\u002F\u002Fgit.sr.ht\u002F~rabbits\u002Forca-examples https:\u002F\u002F0b5vr.com\u002Fwavenerd\u002F https:\u002F\u002Fgithub.com\u002Fwrightwriter\u002FBonzomatic-Compute https:\u002F\u002Fhydra.ojack.xyz https:\u002F\u002Festuary.mcmaster.ca\u002F",{"title":2419,"describe":2420,"type":30},"livecoding in opensource: a guide to the resources and communities","Dazzling animations in a concert, procedural music played from maths. \"livecoding\", a nerdy and artsy hobby with a welcoming community worldwide, let's see what opensource options are waiting for you!",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FSCJ33F",{"id":2424,"room":2425,"start":40,"end":663,"language":70,"track":2426,"speakers":2428,"zh":2436,"en":2439,"tags":2440,"uri":2441},"SFVHC9",{"en":67,"zh-hant":10},{"id":479,"name":2427},{"en":481,"zh-hant":481},[2429],{"id":2430,"avatar":2431,"zh":2432,"en":2435},"ZMEB9N","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FZMEB9N_BH6PvuZ.webp",{"name":2433,"bio":2434},"Alex Lennon","Alex Lennon has been working with embedded Linux and OpenEmbedded for at least 15 years (possibly longer, but his memory is shot). He's the founder of Dynamic Devices Ltd and delivered a keynote at Yocto Project Summit 2022.\r\n\r\nOver the past few months, Alex has been experimenting with AI-assisted development workflows for Yocto projects, learning what works, what doesn't, and where the real productivity gains lie. His recent work includes automated hardware testing, remote target debugging, and power optimization workflows - all enhanced by AI collaboration.\r\n\r\nAlex believes the embedded community needs to figure out AI integration together, sharing both the successes and the failures to develop practical best practices.",{"name":2433,"bio":2434},{"title":2437,"describe":2438,"type":30},"Agentic AI and Rust on Linux: Accelerating Jaguar E-Ink Bring-Up with SSH, Webcams, and Yocto","This session is about **speeding up embedded bring-up** for the **Jaguar EInk Display**: **colour e-paper** on **NXP i.MX93 SOC** with **Yocto \u002F OpenEmbedded** customisation, plus a **Zephyr-based companion chip** for **power management**—targeting **5+ years** operation with the **PMU** approach for realistic **scheduled static signage** workloads.\r\n\r\nThe Linux side display and controller integration are implemented as a **Rust userspace application** on the Linux image—code people often call a “driver,” but here it is **userspace engineering**, iterated like any other critical service.\r\n\r\nThe focus is **agentic AI** (e.g. **Cursor-class agents**) and how it changes the loop when you feed it **real hardware feedback**. I will describe a workflow we relied on: **SSH** to boards (including remote lab setups), **automated deploy** of rebuilt artefacts, and **webcam capture** of what is actually on the panel—fed back into the agent loop so “what we think we programmed” is checked against **what appears on hardware**. That closes the gap between compile-only optimism and **hardware truth**.\r\n\r\nI will cover **partitioning** between **Linux + Rust app** and **Zephyr on the PMU** at an architectural level (**rail sequencing, sleep\u002Fwake, responsibilities**). **PMU application firmware remains closed source**; I will not present proprietary files—only the **integration surface**, **debug strategy**, and lessons that remain useful when part of the stack is private.\r\n\r\nThe product direction includes **long-life maintenance** thinking aligned with **UK-CE RED** and **EU CRA** expectations around **secure boot** and **OTA**—discussed **technically**, not as legal advice.\r\n\r\n### Open artefacts\r\n- [meta-dynamicdevices](https:\u002F\u002Fgithub.com\u002FDynamicDevices\u002Fmeta-dynamicdevices)\r\n- [eink-power-cli](https:\u002F\u002Fgithub.com\u002FDynamicDevices\u002Feink-power-cli)\r\n- [esp-eink](https:\u002F\u002Fgithub.com\u002FDynamicDevices\u002Fesp-eink) (older open e-paper reference material)\r\n\r\n### Audience\r\n\r\nEmbedded \u002F Linux developers; Rust curiosity helps; deep kernel internals not required.\r\n\r\n### Outline (30 minutes)\r\n1. Goals and architecture (**Linux + eink display rust app + Zephyr PMU**)\r\n2. Yocto customisation highlights using Foundries.io stack for secure boot and OTAU\r\n3. Agentic workflow (**SSH**, deploy, logs)\r\n4. Webcam-in-the-loop verification for panel outcomes\r\n5. Failures and guardrails; Q&A (**Open-EP** collaboration themes)\r\n\r\n### Takeaways\r\n\r\n- A repeatable pattern for **remote hardware feedback loops** with agents  \r\n- Pointers to **public repos**\r\n\r\n### Related\r\n- Prior talk (agentic Yocto-era workflows): [Yocto Project Summit 2025.12](https:\u002F\u002Fpretalx.com\u002Fyocto-project-summit-2025-12\u002Ftalk\u002F3KKPQM\u002F)\r\n- Product framing: [etablone](https:\u002F\u002Fetablone.com\u002F)",{"title":2437,"describe":2438,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FSFVHC9",{"id":2443,"room":2444,"start":122,"end":216,"language":96,"track":2445,"speakers":2447,"zh":2455,"en":2458,"tags":2461,"uri":2462},"SJYHWM",{"en":93,"zh-hant":10},{"id":98,"name":2446},{"en":100,"zh-hant":101},[2448],{"id":2449,"avatar":2450,"zh":2451,"en":2454},"QW9GMA","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FQW9GMA_MtYNV24.webp",{"name":2452,"bio":2453},"李盈瑩（Kate)","女性工程師，以 Kate 之名出沒於科技社群，喜歡設計程式、解難題、看小說和塗鴉。\r\n程式主攻 Android App 領域，但不受限於此，樂於吸收新知、挑戰問題以及參與科技社群。\r\n\r\n以[《老姐好像要用 Kotlin 寫專案，能撐30天嗎？》](https:\u002F\u002Fwww.tenlong.com.tw\u002Fproducts\u002F9789864348978)獲得第 12 屆 iT 邦幫忙鐵人賽佳作並出書。\r\n曾在 Kotlin 讀書會 2020 擔任第一梯次導讀和第二梯次線上雙主持。\r\n曾在COSCUP 開源人年會 2021 擔任[講者](https:\u002F\u002Fcoscup.org\u002F2021\u002Fzh-TW\u002Fsession\u002FKEHPVU)。\r\n曾在2022 Kotlin 中文 開發者大會擔任[講者](https:\u002F\u002Fwww.bilibili.com\u002Fvideo\u002FBV1QD4y1s7Lz\u002F?spm_id_from=333.788.recommend_more_video.11)。\r\n從 2019 年口罩議題開始參與 g0v 台灣零時政府社群的腦力激盪，簡訊實聯制的發想人之一。\r\n不定時[寫一些程式相關文章](https:\u002F\u002Fvocus.cc\u002Fuser\u002F@kate)。",{"name":2452,"bio":2453},{"title":2456,"describe":2457,"type":30},"入職第八天，我開始為 AI 寫文件","入職第五天，我拿到了 Claude Code ，開始試著和 AI 同事協作。\r\n第八天，我決定為 AI 寫文件——不是以往的 README.md，是專門餵給 AI 讀的 LLM.md。\r\n第十四天，我把第一個 PR 送出去，同事一看大量的新增檔案，當場傻眼。沒想到幾天後換我審他的 PR 的時候，發現他也開始寫 LLM.md，不只如此，他還寫了一份「文件的文件」，把整套做法立成規範。\r\n這場分享，想講的就是這六十天裡發生的真實故事，以及我們意外發展出的「雙軌文件」玩法：\r\n一份 README 給人類看，另一份 LLM.md 給 AI 看。兩份文件重視的內容完全不一樣。\r\n從這個起點出發，我們一路延伸到架構規範、流程文件、設計 token 系統，慢慢拼出一套團隊能複製的方法論。\r\n\r\n我會分享以下幾件事：\r\n- 為什麼 AI 需要跟人類不一樣的文件？\r\n- 雙軌文件怎麼設計、怎麼演化？\r\n- 個人小習慣是如何變成團隊規範的？\r\n- 以及這套做法真正的代價與限制\r\n\r\n實踐來自一個 Kotlin \u002F Android 團隊，但核心概念適用於任何 JVM 專案。\r\n\r\n無論你是正準備把 AI 工具引入團隊，還是單純對「人跟 AI 一起寫程式」時的工程實踐感興趣，都歡迎來聽聽這段狂踩油門的經歷。",{"title":2459,"describe":2460,"type":30},"Day 8: I Started Writing Documentation for My AI Teammate","On day 5 of my new job, I got Claude Code. On day 8, I started writing documentation specifically for AI — not README, but LLM.md. On day 14, my first PR landed and surprised my teammate, but two weeks later he was writing his own LLM.md, and even drafted a \"documentation of documentation\" to standardize our practice.\r\nThis talk shares the 60-day story of how a Kotlin \u002F Android team developed a dual-track documentation system: README for humans, LLM.md for AI. We'll cover why AI needs different documentation, how the system evolved, how individual practice became team norms, and the real costs and limitations of this approach.\r\nThe practice comes from an Android team, but the concepts apply to any JVM project.",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FSJYHWM",{"id":2464,"room":2465,"start":193,"end":1824,"language":42,"track":2466,"speakers":2468,"zh":2476,"en":2479,"tags":2481,"uri":2482},"SKHXGF",{"en":922,"zh-hant":10},{"id":946,"name":2467},{"en":948,"zh-hant":949},[2469],{"id":2470,"avatar":2471,"zh":2472,"en":2475},"UHZPM3","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FUHZPM3_XVaKy1n.webp",{"name":2473,"bio":2474},"Albert Lu","近乎零停機維護 (Near-Zero Downtime Maintenance)\r\n傳統痛點： 即使是小版本更新（Minor Upgrade），傳統託管服務也需要重啟資料庫服務，這意味著至少會有幾十秒到幾分鐘的斷線（Downtime），連線會被強行中斷。\r\n\r\nOCI 的作法： 透過「熱升級與資料同步」技術。OCI 會在背後悄悄拉起一個運行新版本的全新資料庫實例（Instance），然後利用 Binary Log 自動將舊庫的即時資料同步過去。在切換的「黃金一瞬間」，舊庫維持可讀寫，系統極速將網路流量（DNS\u002FLoad Balancer）導向新庫。應用程式可能只會感覺到一次極其短暫的網絡延遲（甚至小於 1 秒）或少數幾個連線需要重連，而不是整段時間斷線。",{"name":2473,"bio":2474},{"title":2477,"describe":2478,"type":59},"【維運生存戰】迎戰 MySQL 8.0 EOL，我們該如何利用 OCI 特性進行「無感升級」？","OCI 機制的信任度： OCI 新推出的「兩階段控制升級」與「近乎零停機維護」，在營運環境（Production）實際開刀時，大家敢不敢完全信任它的自動化？",{"title":2480,"describe":2478,"type":59},"[Operations Survival Battle] Facing the MySQL 8.0 EOL, how can we leverage OCI features for a \"seamless upgrade\"?",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FSKHXGF",{"id":2484,"room":2485,"start":12,"end":382,"language":96,"track":2486,"speakers":2488,"zh":2496,"en":2499,"tags":2502,"uri":2503},"SMQXRT",{"en":899,"zh-hant":10},{"id":901,"name":2487},{"en":903,"zh-hant":904},[2489],{"id":2490,"avatar":2491,"zh":2492,"en":2495},"JQBE3M","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FJQBE3M_9lP03ek.webp",{"name":2493,"bio":2494},"許貽昇","現任職於奧義智慧科技（CyCraft）資安研究員。研究方向涵蓋靜態程式分析、IOT 安全、LLM 應用於資安自動化等領域。\r\n\r\nSecurity researcher at CyCraft Technology. Research interests include static program analysis, IoT security, and applying LLM to security automation.",{"name":2493,"bio":2494},{"title":2497,"describe":2498,"type":30},"用 CPG 與 LLM 把靜態掃描工具的誤報砍掉——開源混合式漏洞驗證框架實戰","靜態應用安全測試（SAST）工具是軟體安全開發流程中不可或缺的一環，然而許多靜態分析工具往往產生大量誤報，導致開發者在人工審查上耗費巨大成本，甚至因「告警疲勞」而忽略真正的漏洞。\r\n\r\n本議程將分享一套完全開源的混合式漏洞驗證框架，透過三階段流水線大幅降低 SAST 工具的誤報率：\r\n\r\n第一階段：多工具統一掃描。整合 cppcheck、flawfinder、RATS、CodeQL 四款開源 SAST 工具，對 C\u002FC++ 專案進行掃描，並將各工具不同格式的掃描結果統一轉換為標準化格式。\r\n\r\n第二階段：CPG 語義上下文擷取。利用開源程式分析平台 Joern 建構 Code Property Graph（CPG），針對每一筆 SAST 發現的可疑漏洞行，自動擷取完整的函式上下文、資料流路徑、caller\u002Fcallee 關係、以及全域變數賦值分析，為 LLM 提供充分的語義脈絡。\r\n\r\n第三階段：LLM 漏洞驗證。將 CPG 擷取的上下文資訊，交給 LLM 進行漏洞路徑分析，最終對每筆發現做出「確認漏洞」或「誤報」的判定。\r\n\r\n以 SQLite 等真實開源專案為實驗基準，評估了此框架在 Precision、Recall、F1-score 等指標上的表現。實驗結果顯示，此框架能有效過濾大量誤報，同時維持對真實漏洞的高偵測率。\r\n\r\n聽眾將了解如何運用 CPG 為 LLM 提供精準的程式語義上下文的實作經驗，並在真實 C\u002FC++ 專案上降低 SAST 誤報率的具體數據與最佳實踐。",{"title":2500,"describe":2501,"type":30},"Slashing SAST False Positives with CPG + LLM: An Open-Source Hybrid Vulnerability Validation Framework","SAST tools are essential in secure development, yet they generate overwhelming false positives that exhaust developer resources. This talk presents a fully open-source hybrid vulnerability validation framework using a three-stage pipeline: (1) integrating four SAST tools (cppcheck, flawfinder, RATS, CodeQL) with unified output normalization, (2) leveraging Joern to construct Code Property Graphs for extracting function context, dataflow paths, and caller\u002Fcallee relationships, and (3) feeding CPG-enriched context to an LLM for systematic vulnerability path analysis. Evaluated on real-world projects like SQLite, the framework effectively filters false positives while maintaining high true vulnerability detection rates.",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FSMQXRT",{"id":2505,"room":2506,"start":217,"end":1278,"language":328,"track":2507,"speakers":2509,"zh":2517,"en":2520,"tags":2523,"uri":2524},"SVZZQG",{"en":167,"zh-hant":10},{"id":169,"name":2508},{"en":171,"zh-hant":172},[2510],{"id":2511,"avatar":2512,"zh":2513,"en":2516},"KWSYWJ","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FKWSYWJ_sh4wHup.jpg",{"name":2514,"bio":2515},"Tommy Han","Tommy is a software engineer in Hong Kong who loves to work on open source projects in different areas, including toolchains, compilers, apps, and IDE development.",{"name":2514,"bio":2515},{"title":2518,"describe":2519,"type":30},"From PyTorch to Hardware: An Introduction to AI Compilers","AI models may start in frameworks like PyTorch, but getting them to run efficiently on CPUs, GPUs, and specialized accelerators increasingly depends on compiler infrastructure.\r\n\r\nAs machine learning systems have grown more complex and hardware has become more specialized, the compiler layer has become a key bridge between model code and machine execution.\r\n\r\nThis talk introduces AI compilers from a practical perspective, covering why they matter, what kinds of problems they solve, and how open-source projects such as MLIR and OpenXLA\u002FXLA fit into the broader stack. It will also introduce a bit about micrograd and tinygrad.\r\n\r\nIt also explains how ideas such as intermediate representations, lowering, and backend targeting connect frameworks like PyTorch to real hardware.\r\n\r\nA small demo based on PyTorch and Tinygrad helps make the overall pipeline more concrete and easier to understand.",{"title":2521,"describe":2522,"type":30},"從 PyTorch 到硬體：wait what? 甚麼是AI 編譯器...?","現代 AI 系統已不再只運行在單一類型的硬體上，而一個以 PyTorch 撰寫的模型，要如何有效率地在 CPU、GPU 或各種專用加速器上執行，背後越來越依賴編譯器。\r\n\r\n這場分享會從軟體工程的角度，介紹什麼是 AI 編譯器，以及它為什麼變得越來越重要。",[402],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FSVZZQG",{"id":2526,"room":2527,"start":238,"end":239,"language":96,"track":2528,"speakers":2530,"zh":2538,"en":2541,"tags":2544,"uri":2545},"SXDCSE",{"en":39,"zh-hant":10},{"id":384,"name":2529},{"en":386,"zh-hant":10},[2531],{"id":2532,"avatar":2533,"zh":2534,"en":2537},"38ERFR","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002F90109220_3462512763764525_1688018633768304640_n_1_6XK5M24.jpg",{"name":2535,"bio":2536},"Williams Lai","GM! Williams Lai 是在區塊鏈世界打滾了、玩樂了八年的小書僮。目前任職於 Impossible Finance 擔任研究員和 OPs lead ，先前曾是公鏈 Nervos 的佈道者。\r\n\r\nWilliams 非常熱衷於研究任何區塊鏈中的機制設計，從公鏈與 Layer2,以及跨鏈基礎設施，到 Ｗeb3 ,DeFi 和GameFi 等應用領域，只要存在機制設計的地方，就是本人感興趣的範疇。很高興能夠第六次來投稿 Coscup，也很榮幸前五年能夠擔任 Coscup 的講者，並且從中自己得到了許多學習和分享的樂趣。\r\n\r\n今年熱衷的區塊鏈領域的研究題目是 AI 以及支付",{"name":2535,"bio":2536},{"title":2539,"describe":2540,"type":30},"AI Agent 的信任堆疊：鏈上支付、聲譽系統與可驗證計算 - 從 x402、MPP 到 ERC-8004 與 Eigencloud-style verification","AI 正在從被動的助理，逐漸變成能夠執行任務、呼叫 API、甚至與其他 agent 協調工作。但如果 agent 要真正進入開放網路與經濟系統，它們需要的不只是更強的模型，而是一套開放且可驗證的經濟基礎設施，也就是本次演講的主題：鏈上支付、身份聲譽，以及可驗證計算。\r\n\r\n本場演講將探討區塊鏈與開放網路協議如何支援正在形成的 agent economy，筆者認為區塊鏈本身就是一套與 AI 是配的經濟系統，因此本次演講，我會先從 agent payment protocols 談起，包括 Coinbase 推出的x402 與 Tempo \u002F Stripe  共同推出的 Machine Payments Protocol，我會介紹這些協議如何把支付直接嵌入網路請求流程，使 agent 能夠為 API、資料、算力與服務進行機器原生的支付。\r\n另外，筆者也會介紹 ERC-8004 作為 agent identity、reputation 與 validation 的提案標準，並討論 agent 如何在開放網路中被發現、被支付、被評價與被追責。\r\n\r\n最後，我會進一步討論一個目前還未形成市場需求，但將隨著 AI 的發展會有越大的探討空間的技術議題，也就是開源的可驗證計算，在此筆者將會介紹 Eigencloud 與 Brevis 等鏈上可驗證計算項目，去探討EigenAI、EigenCompute、TEE 與 restaking-based validation systems 等可驗證 AI 基礎設施，看看 deterministic inference、verifiable execution 與經濟懲罰機制，如何讓agent 的高風險或者金融行為變得更可信。",{"title":2542,"describe":2543,"type":30},"The Trust Stack for AI Agents: Onchain Payments, Reputation Systems, and Verifiable Computation — From x402 and MPP to ERC-8004 and EigenCloud-Style Verification","AI is gradually evolving from a passive assistant into an autonomous actor capable of executing tasks, calling APIs, and coordinating with other agents. However, for AI agents to meaningfully enter the open internet and broader economic systems, stronger models alone are not enough. They need an open, verifiable economic infrastructure layer. This talk focuses on that emerging trust stack: onchain payments, identity and reputation systems, and verifiable computation.\r\n\r\nThis session will explore how blockchain and open web protocols can support the formation of the agent economy. I argue that blockchain is a naturally compatible economic system for AI agents because it provides programmable settlement, open access, persistent identity, and verifiable execution environments.\r\n\r\nThe talk will begin with agent payment protocols, including Coinbase’s x402 and the Machine Payments Protocol jointly introduced by Tempo and Stripe. I will explain how these protocols embed payment directly into web request flows, enabling agents to make machine-native payments for APIs, data, compute, and services.\r\n\r\nNext, I will introduce ERC-8004 as a proposed standard for agent identity, reputation, and validation. This section will discuss how agents can be discovered, paid, evaluated, and held accountable in an open network environment.\r\n\r\nFinally, I will explore a technical area that has not yet reached full market demand, but will become increasingly important as AI systems take on higher-stakes roles: open and verifiable computation. This section will cover projects such as EigenCloud and Brevis, and examine infrastructure concepts including EigenAI, EigenCompute, TEEs, and restaking-based validation systems. The goal is to explore how deterministic inference, verifiable execution, and economic slashing mechanisms can make high-risk or financial actions performed by AI agents more trustworthy.",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FSXDCSE",{"id":2547,"room":2548,"start":2549,"end":306,"language":70,"track":2550,"speakers":2552,"zh":2556,"en":2559,"tags":2560,"uri":2561},"SXP9SP",{"en":282,"zh-hant":10},"2026-08-08T12:15:00+08:00",{"id":286,"name":2551},{"en":288,"zh-hant":288},[2553],{"id":1566,"avatar":1567,"zh":2554,"en":2555},{"name":1569,"bio":1570},{"name":1569,"bio":1570},{"title":2557,"describe":2558,"type":30},"How OpenEverest Transformed From Single Vendor to Vendor Neutral to CNCF Sandbox Project","OpenEverest is an open-source, cloud-native platform designed to simplify running, managing, and scaling diverse data workloads on Kubernetes. It evolved from a single-vendor initiative to a CNCF Sandbox project in March 2026 by moving beyond its initial vendor-specific platform. While initial support was limited to PostgreSQL, MySQL, and MongoDB due to its single-vendor roots, the project is now making it easy to extend the original system with a modular plugin architecture to enable seamless multi-vendor integration.\r\n\r\nThis transition from a vendor-specific ecosystem to an extensible provider-based design serves as the primary enabler for the community to extend the capability of OpenEverest seamlessly, allowing for the integration of any technology regardless of the founding vendor's roadmap. This session examines the technical refactoring into a plugin-based system and the milestones required to achieve true vendor neutrality and CNCF Sandbox status. Participants will gain a practical blueprint for transforming single-vendor projects into inclusive, community-driven cloud-native platforms.",{"title":2557,"describe":2558,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FSXP9SP",{"id":2563,"room":2564,"start":1770,"end":1175,"language":96,"track":2565,"speakers":2567,"zh":2575,"en":2578,"tags":2581,"uri":2582},"T37ECE",{"en":1253,"zh-hant":10},{"id":901,"name":2566},{"en":903,"zh-hant":904},[2568],{"id":2569,"avatar":2570,"zh":2571,"en":2574},"DP8WEE","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FDP8WEE_5K1qb2X.webp",{"name":2572,"bio":2573},"Jie_Chung (G36maid)","我是 Jie Chung，社群常用 ID 為 G36_maid，目前就讀於國立臺灣師範大學（NTNU）\r\n熱衷於系統程式（System Software）與底層技術開發。專長涵蓋作業系統、虛擬化架構、自架服務 (self-hosting)、資訊安全，以及打造能提升開發體驗的開源工具。\r\n\r\n對開源文化抱持極大熱忱，致力於效能最佳化、現代工具鏈的建構與跨平台系統的探索。\r\n重度 Rust 愛好者與 Zed 編輯器的貢獻者，首屆 Zed Guild 成員。",{"name":2572,"bio":2573},{"title":2576,"describe":2577,"type":30},"編輯器的歷史、現況與未來：從 Electron 到 Rust 與 GPU 驅動的現代化架構 (以 Zed 為例)","程式編輯器是開發者每天接觸最核心的工具。過去十年間，Atom 的誕生將 Web 前端技術（DOM\u002FJS）帶入桌面應用，催生了 Electron 框架，並奠定了 VS Code 的統治地位，但也帶來了效能與記憶體消耗的隱憂。\r\n\r\n本議程將帶領聽眾回顧這段第二次「編輯器之戰」的歷史。我們將首先剖析 Atom 的技術架構與隕落原因，並探討 VS Code 如何透過開源與擴充套件生態贏得開發者喜愛。接著，我們將焦點轉向 Atom 核心團隊重頭打造的開源編輯器——Zed。\r\n\r\n議程將深入解析 Zed 如何捨棄沉重的 Web 框架，運用 Rust 語言與自主研發的 GPUI 框架，將 UI 元件直接交由 GPU 渲染，實現 120 FPS 的極致效能。此外，我們也會探討在 AI 時代下，現代編輯器如何原生整合 MCP\u002FACP 協定，以及講者參與 Zed 開源社群的實務經驗。無論你是工具控、Rust 愛好者，還是對底層架構感興趣的工程師，都能從中獲得技術啟發。",{"title":2579,"describe":2580,"type":30},"The History, Present, and Future of Code Editors: From Electron to Rust and GPU-Driven Modern Architecture (A Case Study of Zed)","Code editors are the most essential tools developers interact with daily. Over the past decade, the birth of Atom brought web frontend technologies (DOM\u002FJS) to desktop applications, spawning the Electron framework and establishing VS Code's dominance. However, this also introduced hidden concerns regarding performance and memory consumption.\r\n\r\nThis session will take the audience through the history of this second \"Editor War.\" We will first analyze Atom's technical architecture and the reasons for its downfall, exploring how VS Code won developers over through its robust extension ecosystem. Next, we will shift our focus to Zed—an open-source editor rebuilt from the ground up by Atom's core team.\r\n\r\nThe session will deep dive into how Zed abandons heavy web frameworks, utilizing Rust and its in-house GPUI framework to render UI components directly via the GPU, achieving an ultimate typing experience at 120 FPS. Furthermore, we will explore how modern editors natively integrate MCP\u002FACP protocols in the AI era, along with practical insights from the speaker's active contributions to the Zed open-source community. Whether you are a tooling enthusiast, a Rust developer, or an engineer interested in low-level architecture, you will find valuable technical inspiration in this talk.",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FT37ECE",{"id":2584,"room":2585,"start":499,"end":500,"language":70,"track":2586,"speakers":2588,"zh":2596,"en":2599,"tags":2600,"uri":2601},"T8AFET",{"en":93,"zh-hant":10},{"id":98,"name":2587},{"en":100,"zh-hant":101},[2589],{"id":2590,"avatar":2591,"zh":2592,"en":2595},"KFSNTA","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002Fmatthias-geisler_L8fGtLr.jpg",{"name":2593,"bio":2594},"Matthias Geisler","True believer in (Kotlin) Multiplatform and working with it since 2019. Builds solutions for Android.\r\nOrganizer of KUG Berlin, GDG Android Berlin, XTC Berlin and more.",{"name":2593,"bio":2594},{"title":2597,"describe":2598,"type":30},"Cycling on KMP","How you combine a Software Engineering and cycling? Most people will now think right away Strava or Garmin. Massive ecosystems, millions of users. But what if we step back and build our own small cycling companion from scratch? What can we learn along the way — about APIs, maps, GPS data and make an integrated platform via KMP? Where will we suffer if we dip into a full of special file formats and integration points and map data. Which concepts can we utilize to have a platform we can expand to something bigger? And how we can make that about community rather than competition. \r\nWhether you're a cyclist curious about code or a developer curious about cycling, let’s hope on and get started.",{"title":2597,"describe":2598,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FT8AFET",{"id":2603,"room":2604,"start":1112,"end":454,"language":96,"track":2605,"speakers":2607,"zh":2615,"en":2618,"tags":2621,"uri":2622},"TEDPF3",{"en":522,"zh-hant":10},{"id":524,"name":2606},{"en":526,"zh-hant":10},[2608],{"id":2609,"avatar":2610,"zh":2611,"en":2614},"J8QABA","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FJ8QABA_xCpxmAK.jpg",{"name":2612,"bio":2613},"alanhc","目前在 IC 廠從事 Android 產品 系統整合與開發，過去曾在系統廠擔任 BMC Engineer，負責資料中心伺服器平台的 BMC firmware、OpenBMC、與系統開發及除錯。\r\n\r\n我專注於 embedded Linux、firmware–software integration、Performance、系統級 debugging，對 open source、firmware–software co-design，以及可靠且可擴展的現代運算基礎設施充滿熱情。\r\n\r\n希望成為能聽懂硬體、理解韌體、看透系統行為的工程師——嵌入式系統的「通靈王」。\r\n\r\nWebsite: https:\u002F\u002Falanhc.github.io\u002F",{"name":2612,"bio":2613},{"title":2616,"describe":2617,"type":30},"韌體開發與 Agentic AI：讓 AI 跑 Build，你來解難題","Embedded \u002F Firmware 工程師的日常充滿大量重複性工作——fetch code、sync code、build、驗證、燒錄、收 log。這些流程每個 project 都要重來一遍，耗時卻幾乎不需要人的判斷力。\r\n\r\n本議程分享如何將這套 routine 封裝成 AI Agent Skills，讓 AI 自動執行整個 build pipeline，在真正需要人介入時（編譯錯誤、版本衝突、硬體異常）才發出通知。透過這個架構，一個工程師可以同時推進多個 project，把時間和精力留給真正需要思考的部分。\r\n\r\n以 AOSP \u002F Pixel 開發流程為具體範例，展示從 repo sync、vendor blob 驗證、版本對齊到 fastboot flashall 的完整 pipeline 如何被 AI Agent 接管，以及如何設計 skill 的介入邊界——哪些交給 AI，哪些留給人。",{"title":2619,"describe":2620,"type":30},"Firmware Development and Agentic AI","Embedded and Firmware engineers spend a significant portion of their day on repetitive tasks — fetching code, syncing, building, verifying, flashing, and collecting logs. These steps repeat across every project, consuming time while requiring almost no human judgment.\r\n\r\nThis talk shares how to encapsulate these routines into AI Agent Skills, allowing AI to autonomously execute the entire build pipeline and only notify the engineer when genuine intervention is needed — compile errors, version conflicts, or hardware anomalies. With this approach, a single engineer can drive multiple projects in parallel, reserving their time and focus for work that actually requires human thinking.\r\n\r\nUsing the AOSP \u002F Pixel development workflow as a concrete example, the talk demonstrates how a complete pipeline — from repo sync, vendor blob verification, and build ID alignment, all the way to fastboot flashall — can be handed off to an AI Agent. It also covers how to design the intervention boundary: what to delegate to AI, and what to keep in human hands.",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FTEDPF3",{"id":2624,"room":2625,"start":239,"end":408,"language":96,"track":2626,"speakers":2628,"zh":2641,"en":2644,"tags":2646,"uri":2647},"TG89GH",{"en":350,"zh-hant":10},{"id":352,"name":2627},{"en":354,"zh-hant":355},[2629,2636],{"id":2630,"avatar":2631,"zh":2632,"en":2635},"LTQYEY","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FLTQYEY_ZmhOFIh.webp",{"name":2633,"bio":2634},"weichen","人類學背景，關注數位議題與網路治理\r\ngithub.com\u002Fwcl-dev\r\n研究筆記：wcl-dev.github.io\u002Fresearch-site\r\n網路運作科普：wcl-dev.github.io\u002FHIW",{"name":2633,"bio":2634},{"id":2637,"avatar":3,"zh":2638,"en":2640},"8CLJYF",{"name":2639,"bio":10},"kevin",{"name":2639,"bio":10},{"title":2642,"describe":2643,"type":30},"從假訊息到資訊操弄：以案例理解 FIMI 與認知韌性","本場分享將以「無背景友善」為原則，帶領參與者理解 FIMI（Foreign Information Manipulation and Interference，外國資訊操弄與干預）的基本概念，並以無邊界集團相關公開案例作為切點，說明資訊操弄如何不只存在於政治新聞或明顯假訊息中，也可能嵌入日常生活粉專、內容農場、導流網站與平台互動機制之中。\r\n\r\n分享將先說明為何 FIMI 的重點不只是內容真假，而是行為者、協調模式、基礎設施與操作目的。接著以無邊界集團為例，介紹如何從粉專透明度、貼文模式、留言導流、內容農場網絡與跨境管理跡象，觀察一套資訊操弄操作如何被建立、維持與變現。\r\n\r\n最後，也會分享調查工具打造與使用經驗，包括如何記錄粉絲專頁與網站變化、保存證據、追蹤 URL 與內容農場關聯，以及如何把零散線索整理成可被討論、查證與協作的調查材料。希望讓參與者即使沒有相關背景，也能理解網路與認知韌性，並帶走一套可入門實作的觀察方法。",{"title":2645,"describe":2643,"type":30},"From Disinformation to Influence Operations: Understanding FIMI through the Borderless Group Case",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FTG89GH",{"id":2649,"room":2650,"start":12,"end":382,"language":70,"track":2651,"speakers":2653,"zh":2661,"en":2664,"tags":2665,"uri":2666},"THXUNH",{"en":282,"zh-hant":10},{"id":286,"name":2652},{"en":288,"zh-hant":288},[2654],{"id":2655,"avatar":2656,"zh":2657,"en":2660},"MVTE9V","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FMVTE9V_2QZ4rkU.webp",{"name":2658,"bio":2659},"梶山 隆輔 \u002F KAJIYAMA, Ryusuke","KAJIYAMA, Ryusuke (梶山 隆輔) is Open Source Specialist with 20+ years of experience of systems design using MySQL and open source solutions. He originally joined MySQL AB, and is at Oracle through acquisitions. He was managing MySQL Solution Engineering team of Asia Pacific region at Oracle and currently responsible for Business Development role of Oracle's open source database products including MySQL, PostgreSQL, OpenSearch, Redis, and Valkey.",{"name":2658,"bio":2659},{"title":2662,"describe":2663,"type":30},"PostgreSQL on Oracle Cloud: Open Source Databases, Managed Infrastructure, and the AI-Native Application Stack","Oracle is known for Oracle Database, one of the most widely used proprietary database. So why is Oracle offering a managed PostgreSQL service?\r\n\r\nThis session starts with that unexpected question and uses it as a lens to explore a broader shift in modern application architecture: enterprises increasingly want the openness, ecosystem, and developer familiarity of PostgreSQL, while also needing cloud-native operations, high availability, scalable infrastructure, and integration with AI services.\r\n\r\nOCI Database with PostgreSQL is Oracle Cloud Infrastructure’s fully managed PostgreSQL service. Rather than positioning it as a replacement for self-managed PostgreSQL or as a proprietary database story, this talk examines the technical design choices behind running PostgreSQL as a managed service: separation of compute and storage, database-optimized shared storage, automated backup and restore, read scaling through replicas, high availability with fast failover, and operational integration with cloud monitoring and security services.\r\n\r\nA key focus will be PostgreSQL’s extension ecosystem and why it matters for AI-era applications. We will look at extensions such as pgvector for vector search, PostGIS for geospatial workloads, pglogical for replication scenarios, and other operational extensions that make PostgreSQL more than a traditional relational database. In particular, the session will explain how PostgreSQL can act as a vector store for retrieval-augmented generation (RAG), hybrid search, and semantic search use cases, while remaining part of a familiar SQL-based application architecture.\r\n\r\nThe talk will also cover practical architecture patterns: using PostgreSQL with object storage, streaming, AI services, application containers, and search\u002Fcache services to build modern applications on open source components. The goal is not to present a product pitch, but to discuss what it means to operate an open source database as an enterprise-grade managed service, and where the trade-offs appear between openness, operational control, performance, resilience, and cloud integration.\r\n\r\nAttendees will leave with a clearer understanding of how managed PostgreSQL services are evolving, how PostgreSQL fits into AI application architectures, and why even a company strongly associated with Oracle Database sees PostgreSQL as an important part of the modern open source data platform.",{"title":2662,"describe":2663,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FTHXUNH",{"id":2668,"room":2669,"start":709,"end":68,"language":70,"track":2670,"speakers":2672,"zh":2680,"en":2683,"tags":2684,"uri":2685},"TPZAHJ",{"en":684,"zh-hant":10},{"id":686,"name":2671},{"en":688,"zh-hant":689},[2673],{"id":2674,"avatar":2675,"zh":2676,"en":2679},"JZTZNV","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FJZTZNV_nik8dk1.webp",{"name":2677,"bio":2678},"Brian Su","Brian Su is a developer advocate at Bigstack, a cloud-native infrastructure company that developed CubeCOS, an open-source unified cloud operating system (OS) designed to run virtual machine (VM) and Kubernetes workloads on a single platform. With a background in systems engineering and customer infrastructure, Brian focuses on making complex distributed systems accessible through technical content, documentation, and developer resources.\r\n\r\nBrian is actively involved in the CubeCOS open-source community and contributes to developer onboarding. Brian guides internal security pipeline provisioning, maintains secure code operations, and supports enterprise deployments across hybrid and air-gapped environments.\r\n\r\nTo stay sharp, Brian also maintain a personal homelab to experiment with the latest infrastructure and security tooling.",{"name":2677,"bio":2678},{"title":2681,"describe":2682,"type":30},"Securing the Open Source Telco Cloud: SBOMs, Supply Chains, and Compliance at Scale","Building a sovereign, open-source cloud platform for telecommunications environments is more challenging than it seems. The telco use case involves addressing the need for an evolving security architecture. Once teams commit to SBOM compliance, a long tail of previously invisible vulnerabilities suddenly becomes visible.\r\n\r\nThis talk will demonstrate how utilising an open-source toolkit can help teams establish a solid foundation for compliance and security hardening on telco cloud platforms. Tools such as Syft, Grype, DefectDojo, the OpenSSF Scorecard and GitHub for Open Source can help development and security teams establish an integrated CI\u002FCD pipeline for DevSecOps, gain visibility into the security posture of your project, deliver a verifiable SBOM to end users and automate dependency monitoring and management.\r\n\r\nWe will walk through how component scanning and SBOM management can reduce unresolved vulnerabilities by 80%. Integrating security tooling into the CI\u002FCD pipeline reduces time to resolution and enhances the developer experience. Ingesting the scan findings and artefacts into a centralised platform enables faster triage, prioritisation and resolution of vulnerabilities in a methodological manner. We will also demonstrate how open source tooling feeds into compliance and sovereignty for telco deployments. \r\n\r\nAttendees will learn how to design an SBOM-native release workflow and how to triage an SBOM without stopping delivery. They will also discover why open source and digital sovereignty go hand in hand. They will also learn how to turn security compliance from a bottleneck into a competitive advantage and understand why open source is the only foundation that makes digital sovereignty possible.",{"title":2681,"describe":2682,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FTPZAHJ",{"id":2687,"room":2688,"start":1278,"end":499,"language":96,"track":2689,"speakers":2691,"zh":2695,"en":2698,"tags":2701,"uri":2702},"TRRMY3",{"en":121,"zh-hant":10},{"id":125,"name":2690},{"en":127,"zh-hant":128},[2692],{"id":692,"avatar":693,"zh":2693,"en":2694},{"name":695,"bio":696},{"name":695,"bio":696},{"title":2696,"describe":2697,"type":30},"關於開放原始碼 x 人工智慧，聯合國怎麼說？","本演講將以講者親身與會經驗，系統性分享 UN Open Source Week 2026 中，聯合國和全球專家學者如何看待開放原始碼對於發展人工智慧與新興技術的助益。內容預計包含開放原始碼 × 人工智慧的國際共識、跨境協作治理框架、開源硬體與基礎建設趨勢、以及開放代理（Open Agents）的透明度與問責機制。同時，該大會預計也將為「數位合作的開放性」提出的普惠創新政策路徑——可謂是開放文化在數位時代的集大成。",{"title":2699,"describe":2700,"type":30},"What did the UN say about Open Source x AI?","This talk draws on the speaker's firsthand experience at UN Open Source Week 2026 to offer a systematic account of how the United Nations and global experts view the role of open source in advancing artificial intelligence and emerging technologies. Topics will include the international consensus on open source × AI, cross-border collaborative governance frameworks, trends in open source hardware and infrastructure, and transparency and accountability mechanisms for Open Agents. The event is also expected to put forward inclusive innovation policy pathways under the theme of \"Openness for Digital Cooperation\" — making it a landmark gathering for open culture in the digital age.",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FTRRMY3",{"id":2704,"room":2705,"start":500,"end":11,"language":42,"track":2706,"speakers":2708,"zh":2715,"en":2718,"tags":2721,"uri":2722},"TTP37G",{"en":9,"zh-hant":10},{"id":15,"name":2707},{"en":17,"zh-hant":18},[2709],{"id":2710,"avatar":2711,"zh":2712,"en":2714},"EEZE3E","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FDennis-Chen-Rladys-anniversity-photo_wSOvqO7.jpg",{"name":2713,"bio":10},"Dennis Raylin Chen",{"name":2713,"bio":10},{"title":2716,"describe":2717,"type":30},"以 OpenRefine 加上 Python，實作臺灣台語的地名半自動處理","資料科學工具 OpenRefine，加上 Python 的台語處理套件，去半自動化處 理Wikidata 與 OpenStreetMap的地名項目的標籤。另外也嘗試用爬蟲抓維基百科上模板當中的臺灣台語羅馬字拼音方案與白話字的名稱。",{"title":2719,"describe":2720,"type":30},"Using OpenRefine and Python, Implement semi-automatic process of Taiwan Taigi name place","By using data science tool OpenRefine, and Python Taigi package, we can implement a semi-automatic process of Wikidata and OpenStreetMap name place labels. And we will try to write web scrapping to get Taiwan Taigi romaization names or Pehoeji names in the Wikipeida template.",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FTTP37G",{"id":2724,"room":2725,"start":306,"end":12,"language":70,"track":2726,"speakers":2728,"zh":2732,"en":2735,"tags":2736,"uri":2737},"TVPACP",{"en":93,"zh-hant":10},{"id":98,"name":2727},{"en":100,"zh-hant":101},[2729],{"id":2590,"avatar":2591,"zh":2730,"en":2731},{"name":2593,"bio":2594},{"name":2593,"bio":2594},{"title":2733,"describe":2734,"type":30},"Design Patterns 101","What is a Factory? Can a Bridge do something else than connect 2 points? What's the difference between Command and Event? What other patterns are out there?\r\nDesign patterns are the building blocks of elegant software architecture. But when and how to apply them effectively?\r\nLet's discover together the most essential patterns through real-world examples and practical demonstrations that will transform how you approach code design.",{"title":2733,"describe":2734,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FTVPACP",{"id":2739,"room":2740,"start":68,"end":69,"language":96,"track":2741,"speakers":2743,"zh":2751,"en":2754,"tags":2756,"uri":2757},"TYA9AF",{"en":192,"zh-hant":10},{"id":196,"name":2742},{"en":198,"zh-hant":198},[2744],{"id":2745,"avatar":2746,"zh":2747,"en":2750},"S3ZNTK","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FS3ZNTK_16AdkGm.webp",{"name":2748,"bio":2749},"Macpaul Lin","馬克泡曾經參與無線路由器、開發板、以及Android智慧型手機晶片等等的產品開發。現任職於聯發科技，參與IOT與Edge AI開發板的開發套件相關工作。業餘時長期參與開源社群活動。\r\n由於買了半價的gemini每天都焦慮點數用不完怎麼辦，只好想辦法每天固定出三張AI廢圖。",{"name":2748,"bio":2749},{"title":2752,"describe":2753,"type":30},"反正都要抄答案了，為什麼不用AI做？分享用AI工具移植Linux驅動到u-boot上的經驗","在嵌入式系統開發中，將成熟的 Linux Kernel 驅動或框架移植至 U-Boot 是建構硬體支援的重要環節。傳統流程需由熟悉該裝置的工程師，研讀軟硬體規格，人工分析兩者 API 差異、手動裁減中斷處理與併發機制，並重新適配核心結構，必要時還必須升級同步framework版本差異。過程耗時且易出錯。\r\n本講題將分享如何轉變開發方式，透過AI輔助，參照Linux Code生成 U-Boot 兼容代碼。我們將分享如何透過AI工具輔助與人工驗證，加速開發週期。",{"title":2755,"describe":2753,"type":30},"If we’re going to copy the answer anyway, why not let AI do it? Insights into porting Linux drivers to U-Boot using AI tools.",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FTYA9AF",{"id":2759,"room":2760,"start":40,"end":663,"language":70,"track":2761,"speakers":2763,"zh":2771,"en":2774,"tags":2775,"uri":2776},"U8EAUL",{"en":1007,"zh-hant":10},{"id":901,"name":2762},{"en":903,"zh-hant":904},[2764],{"id":2765,"avatar":2766,"zh":2767,"en":2770},"LNVYY7","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FLNVYY7_YjYiuRx.webp",{"name":2768,"bio":2769},"Sailesh Singh","Sailesh Singh is a System and Network Administrator and backend engineer with over 6 years of experience in Linux, cloud infrastructure, and web development. He is the Founder & CEO of Nepbrain Labs, focused on building scalable systems with Django, Kubernetes, and cloud-native technologies. He is an active open-source advocate and co-leads GNOME Nepal and Ubuntu Nepal, contributing to initiatives like Software Freedom Day and UbuCon Asia. He has spoken at international events including GNOME Asia Summit, UbuCon Asia, and gRPC Conf (organized by the Linux Foundation at Google). He is also an organizer of UbuCon Asia 2026 and a foundation member of the GNOME and Ubuntu.",{"name":2768,"bio":2769},{"title":2772,"describe":2773,"type":30},"Running gRPC in Production on Ubuntu: Scaling Microservices with Kubernetes, Observability, and Resilience","Modern distributed systems require high performance, low latency, and reliability at scale. While gRPC provides an efficient communication layer using HTTP\u002F2 and Protocol Buffers, running it in production introduces challenges beyond basic implementation.\r\n\r\nIn this talk, we explore how to design, deploy, and scale gRPC-based microservices on Ubuntu-powered infrastructure. Drawing from real-world experience, the session covers service discovery in Kubernetes environments, observability using tools, and advanced load balancing techniques.\r\n\r\nWe will dive into production-critical topics such as retry strategies, connection management, and backward compatibility, along with deployment strategies including canary releases and rolling updates.\r\n\r\nBy leveraging Ubuntu’s stability, long-term support (LTS), and open-source ecosystem, **attendees will learn how to build reliable and scalable backend systems and avoid common pitfalls that are often not covered in documentation**.",{"title":2772,"describe":2773,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FU8EAUL",{"id":2778,"room":2779,"start":605,"end":476,"language":70,"track":2780,"speakers":2782,"zh":2790,"en":2793,"tags":2796,"uri":2797},"UCQREE",{"en":1253,"zh-hant":10},{"id":901,"name":2781},{"en":903,"zh-hant":904},[2783],{"id":2784,"avatar":2785,"zh":2786,"en":2789},"LEW3DP","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FLEW3DP_KXLydEG.webp",{"name":2787,"bio":2788},"Yuyuan Yuan","Yuyuan Yuan is a Solution Architect at ZettaScale Technology, where he works on Zenoh — an open-source Rust protocol for pub\u002Fsub, queries, and distributed storage from edge to cloud.\r\n\r\nHe works on Hiroz, a pure-Rust Zenoh-native ROS 2 middleware, and holds commit access to both Eclipse Zenoh and ros2\u002Frmw_zenoh — the official Zenoh-based RMW for ROS 2. His work spans protocol design, async runtime architecture, and multi-language bindings from a Rust core (Python via PyO3, Go via C FFI).\r\n\r\nHe has presented at FOSDEM 2026 (Robotics and Simulation devroom), GOSIM, RUSTAsia 2025, and COSCUP 2023 and 2022. When not writing Rust, he explores formal proofs in Lean, runs marathons, and has completed a 226 km Ironman-distance triathlon. He runs NixOS btw.",{"name":2787,"bio":2788},{"title":2791,"describe":2792,"type":30},"From ROS to Rust: Hiroz, Zenoh, and a Robotics Stack You Can `cargo build`","[Eclipse Zenoh](https:\u002F\u002Fzenoh.io) unifies data in motion, data at rest, and computations with a single protocol—from microcontrollers to cloud infrastructure. This open-source Rust stack delivers pub\u002Fsub, geo-distributed queries, and storage with low-latency performance.\r\n\r\n**Hiroz** is a pure-Rust ROS 2 middleware built natively on Zenoh—no C dependencies, just `cargo build`. Wire-compatible with standard ROS 2 nodes via [rmw_zenoh](https:\u002F\u002Fgithub.com\u002Fros2\u002Frmw_zenoh), it provides Rust\u002FGo\u002FPython bindings with zero-copy support and Protobuf serialization. Fully open-source Apache-2.0: [github.com\u002FZettaScaleLabs\u002Fhiroz](https:\u002F\u002Fgithub.com\u002FZettaScaleLabs\u002Fhiroz). Check out the [website](https:\u002F\u002Fzettascalelabs.github.io\u002Fhiroz\u002F)! \r\n\r\nThis talk covers why the two existing Rust-in-ROS-2 approaches each involve real trade-offs—ros2_rust wraps the C RCL layer and stays wire-compatible but requires a full ROS 2 install and spreads `unsafe` FFI throughout; dora-rs goes pure Rust but drops wire compatibility entirely—and how hiroz takes a third path: pure Rust, no C dependencies, and wire-compatible with standard ROS 2 nodes out of the box. We then go inside hiroz's async runtime design, message generation pipeline, and multi-language binding strategies.",{"title":2794,"describe":2795,"type":30},"從 ROS 到 Rust：Hiroz、Zenoh，以及一個用 `cargo build` 就能建置的機器人通訊框架","[Eclipse Zenoh](https:\u002F\u002Fzenoh.io) 以單一協定統一動態資料、靜態資料與運算——從微控制器到雲端基礎設施皆適用。這套開源 Rust 技術棧提供發布／訂閱、地理分散式查詢與儲存，並具備低延遲效能。\r\n\r\n**Hiroz** 是一套以 Zenoh 為基礎、純 Rust 實作的 ROS 2 中介軟體——無需 C 語言相依套件，只需 `cargo build`。透過 [rmw_zenoh](https:\u002F\u002Fgithub.com\u002Fros2\u002Frmw_zenoh) 與標準 ROS 2 節點保持線路相容性，並提供 Rust／Go／Python 語言綁定、零複製支援與 Protobuf 序列化。完全開源，採用 Apache-2.0 授權：[github.com\u002FZettaScaleLabs\u002Fhiroz](https:\u002F\u002Fgithub.com\u002FZettaScaleLabs\u002Fhiroz)。\r\n\r\n本演講將介紹現有兩種 Rust in ROS 2 方案各自的取捨——ros2_rust 封裝 C RCL 層以保持線路相容性，但需要完整的 ROS 2 安裝且 `unsafe` FFI 遍佈各處；dora-rs 走純 Rust 路線，卻放棄了與現有 ROS 2 節點的線路相容性——並說明 hiroz 如何走出第三條路：純 Rust、無 C 相依套件，且與標準 ROS 2 節點開箱即用地保持線路相容。最後深入探討 hiroz 的非同步執行時設計、訊息生成流水線，以及多語言綁定策略的實作細節。",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FUCQREE",{"id":2799,"room":2800,"start":453,"end":94,"language":328,"track":2801,"speakers":2803,"zh":2818,"en":2821,"tags":2824,"uri":2825},"UEUXUA",{"en":167,"zh-hant":10},{"id":169,"name":2802},{"en":171,"zh-hant":172},[2804,2811],{"id":2805,"avatar":2806,"zh":2807,"en":2810},"TB8FK8","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FTB8FK8_gmlbtw3.webp",{"name":2808,"bio":2809},"Mario Behling","Mario Behling is the co-founder of FOSSASIA and a member of the Visdom team. He works on open source projects, developer communities, and open technology events across Asia and beyond. At FOSSASIA, he supports projects such as Visdom, eventyay, PSLab, and other open source tools, with a focus on community collaboration, sustainable development, and practical open source infrastructure.",{"name":2808,"bio":2809},{"id":2812,"avatar":2813,"zh":2814,"en":2817},"XWW9AQ","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002F6468f3df-ff33-4e45-b2e6-f17576f24ddc_fChltVe.jpg",{"name":2815,"bio":2816},"François Cartegnie","I'm here to set up the remote session for the talk.\r\nI belong to another unrelated project.",{"name":2815,"bio":2816},{"title":2819,"describe":2820,"type":30},"From Experiments to Dashboards: Modernizing Visdom for Open AI Workflows \u002F 從實驗到儀表板：為開放 AI 工作流程現代化 Visdom","Visdom is an open source tool for live visualizations and lightweight dashboards in Python, PyTorch, NumPy, and AI workflows. In this session, François Cartegnie will facilitate the presentation and conversation with Mario Behling and the Visdom team which will share the story of Visdom’s modernization at FOSSASIA, including the newest release, local deployment, and an emerging online version that lets researchers and developers quickly use Visdom for their projects.\r\n\r\nVisdom 是一套開源工具，可用於 Python、PyTorch、NumPy 和 AI 工作流程中的即時視覺化與輕量級儀表板。在本場次中，Visdom 團隊的 Mario Behling 將分享 Visdom 在 FOSSASIA 推動現代化的故事，包括最新版本、本機部署，以及正在發展中的線上版本，讓研究人員和開發者能更快速地將 Visdom 用於自己的專案。",{"title":2822,"describe":2823,"type":30},"From Experiments to Dashboards: Modernizing Visdom for Open AI Workflows","AI and scientific computing workflows produce metrics, plots, images, logs, model outputs, and many other experimental results. Developers and researchers need practical ways to inspect this data, compare experiments, and share insights without depending only on closed or heavyweight platforms.\r\n\r\nVisdom is an open source tool for creating, organizing, and sharing live visualizations of rich data. Originally developed at Facebook AI Research and later transitioned to FOSSASIA, Visdom has a long history and an existing user base in Python, PyTorch, NumPy, and scientific computing communities.\r\n\r\nIn this session, François Cartegnie will facilitate the presentation and conversation with Mario Behling and the Visdom team that shares the story of Visdom’s next phase: how the project is being modernized for current AI workflows, what kinds of technical and community challenges come with maintaining an established open source tool, and why open visualization tools remain relevant in today’s AI ecosystem.\r\n\r\nThe talk will cover two important directions for Visdom: local-first deployment for developers who want to run the tool in their own environment, and a new online version that enables researchers and developers to quickly upload or connect their projects and start using Visdom without setting up their own instance first.\r\n\r\nAt COSCUP, we will also present the newest release of Visdom and discuss current development directions, including usability, frontend and backend modernization, Python client improvements, documentation, testing, and new visualization features. The session invites COSCUP participants to explore how open source communities can keep important AI tools useful, sustainable, and open.\r\n\r\nThe session will be held in a mix of English and Chinese, with team members helping with translation to make it accessible to both international and local audiences.",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FUEUXUA",{"id":2827,"room":2828,"start":238,"end":239,"language":96,"track":2829,"speakers":2831,"zh":2839,"en":2842,"tags":2843,"uri":2844},"UFSFQW",{"en":684,"zh-hant":10},{"id":686,"name":2830},{"en":688,"zh-hant":689},[2832],{"id":2833,"avatar":2834,"zh":2835,"en":2838},"NWLNCB","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FNWLNCB_oo92IUg.webp",{"name":2836,"bio":2837},"Marie Gigarel","Secretary General of IMA資訊經理人協會\r\nhttps:\u002F\u002Fima.org.tw\u002F\r\nhttps:\u002F\u002Ftw.linkedin.com\u002Fcompany\u002Fima-tw",{"name":2836,"bio":2837},{"title":2840,"describe":2841,"type":30},"From Code Contributor to Industry Power --How Open Source Becomes Taiwan’s AI Strategy","Taiwan helped build the world’s hardware industry. But in the AI era, competitive advantage is shifting beyond manufacturing toward software ecosystems, developer communities, data, and open-source collaboration.\r\n\r\nToday, the world’s leading AI companies are increasingly built on open ecosystems. NVIDIA transformed CUDA into a developer platform. Hugging Face became a global AI community hub. Projects such as PyTorch, Kubernetes, vLLM, and ROS are now part of the invisible infrastructure powering modern AI innovation.\r\n\r\nThis creates both a challenge and an opportunity for Taiwan.\r\n\r\nWhile Taiwan possesses world-class engineering talent and deep hardware expertise, its open-source ecosystem has long depended mainly on individual passion rather than systematic industry participation. As AI development becomes more open and collaborative, Taiwan must rethink how enterprises, startups, governments, and communities work together to create long-term software influence and industrial value.\r\n\r\nThis session explores how open source is evolving from a “developer activity” into a national AI strategy. The talk will introduce the concept of the “Open Source Flywheel” — where contribution creates influence, influence attracts talent and capital, and commercialization further accelerates ecosystem growth.\r\n\r\nThrough global examples and Taiwan’s emerging AI initiatives, the session will discuss how Taiwan can move beyond the role of hardware supplier and become an active participant in shaping the future AI ecosystem.\r\n\r\nTopics include open-source business models, AI geopolitics, ecosystem competition, and Taiwan’s opportunities in Edge AI, cloud infrastructure, robotics, inference optimization, and multilingual AI.\r\n\r\nAs AI reshapes the global technology landscape, the key question is no longer whether Taiwan can manufacture the world’s hardware — but whether Taiwan can help build the world’s AI ecosystem.",{"title":2840,"describe":2841,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FUFSFQW",{"id":2846,"room":2847,"start":2028,"end":41,"language":70,"track":2848,"speakers":2850,"zh":2857,"en":2860,"tags":2861,"uri":2862},"UPWEEP",{"en":684,"zh-hant":10},{"id":686,"name":2849},{"en":688,"zh-hant":689},[2851,2854],{"id":1283,"avatar":1284,"zh":2852,"en":2853},{"name":1286,"bio":1287},{"name":1286,"bio":1287},{"id":1290,"avatar":1291,"zh":2855,"en":2856},{"name":1293,"bio":1294},{"name":1293,"bio":1294},{"title":2858,"describe":2859,"type":30},"Your Container Images Are a Liability: The Supply Chain Debt Nobody Is Paying Down","Every container you deploy carries debt you didn't write. The average base image ships with 200-400 packages your application never calls, each one a potential CVE, each one expanding the blast radius of a breach. Teams run Trivy or Grype, get a wall of 400 alerts, patch the criticals, suppress the rest, and ship. The scan-patch-suppress cycle creates an illusion of security hygiene while the actual attack surface stays enormous.\r\n\r\nThe good news: the ecosystem is finally pushing back. Google's distroless project has been around for years, Chainguard built a business on minimal images, and Docker Hardened Images went fully open source under Apache 2.0 in late 2025, putting 1,000+ minimal, SBOM-signed images one pull away from every developer. Yet most teams still default to node:latest.\r\n\r\nThis talk dissects why container supply chain debt accumulates and what a different default looks like. Through live audits comparing standard, slim, distroless, and hardened base images for the same application, we'll examine size, CVE counts, and actual runtime dependencies. You'll leave with practical patterns: multi-stage builds done right, automated base image rebuild pipelines, and policy-as-code for image provenance, so minimal becomes the default without slowing teams down.",{"title":2858,"describe":2859,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FUPWEEP",{"id":2864,"room":2865,"start":122,"end":216,"language":96,"track":2866,"speakers":2868,"zh":2874,"en":2877,"tags":2880,"uri":2881},"UQTYUN",{"en":407,"zh-hant":10},{"id":410,"name":2867},{"en":412,"zh-hant":412},[2869],{"id":2870,"avatar":3,"zh":2871,"en":2873},"3DW9L7",{"name":2872,"bio":10},"Vic",{"name":2872,"bio":10},{"title":2875,"describe":2876,"type":30},"用 Signals 重新認識 Angular：寫給新手的現代框架導覽","所有前端框架，其實都在回答同一個問題：資料變了，畫面要怎麼跟著變？過去十幾年，從 React 的 useState、Vue 的 ref 到 Solid 的 signal，各家的答案正悄悄收斂到同一個方向，就是細粒度的響應式。\r\n\r\n而提到 Angular，很多人腦中還停在 NgModule 滿天飛、樣板程式碼又臭又長的印象。但這幾年的 Angular 已經是另一個框架了，而帶起這波轉變的核心，就是 Signals。\r\n\r\n這場演講會用 Signals 當作一面鏡子，先帶你看懂現代前端框架到底在思考什麼：為什麼「資料到畫面」這個看似簡單的問題這麼難、各框架又是怎麼一步步逼近答案的。接著我們會落到實處，實際看 Angular 的 Signals 寫起來長什麼樣子（會帶到 Angular 的各種 Signals 新語法）。\r\n\r\n無論你是好奇 Angular 近況的前端開發者，還是正在做技術選型的人，都希望這場演講能透過 Signals 這個切角，重新認識一個跟你印象中很不一樣的 Angular。",{"title":2878,"describe":2879,"type":30},"Rediscovering Angular Through Signals: A Beginner's Guide to Modern Frontend Frameworks","Every frontend framework is really answering the same question: when data changes, how does the UI keep up? Over the past decade or so, the answers—from React's useState to Vue's ref to Solid's signal—have been quietly converging on the same direction: fine-grained reactivity.\r\n\r\nFor many people, the word \"Angular\" still conjures up an image of NgModules everywhere and endless boilerplate. But Angular has become a different framework in recent years, and the core of that transformation is Signals.\r\n\r\nThis talk uses Signals as a mirror. We'll start by making sense of what modern frontend frameworks are actually thinking about: why \"from data to UI\" is such a deceptively hard problem, and how each framework has inched its way toward an answer. Then we'll get concrete and look at what Angular's Signals actually look like in practice (including the various new Signals syntax Angular has introduced).\r\n\r\nWhether you're a frontend developer curious about where Angular stands today or someone weighing it for a technical decision, this talk aims to help you rediscover—through the lens of Signals—an Angular that's quite different from the one you remember.",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FUQTYUN",{"id":2883,"room":2884,"start":216,"end":217,"language":70,"track":2885,"speakers":2887,"zh":2895,"en":2898,"tags":2899,"uri":2900},"UYUKW7",{"en":1063,"zh-hant":10},{"id":1065,"name":2886},{"en":1067,"zh-hant":1068},[2888],{"id":2889,"avatar":2890,"zh":2891,"en":2894},"PPCWYN","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FKIMG0548_GCqGfFG.JPG",{"name":2892,"bio":2893},"Naruhiko Ogasawara","Naruhiko Ogasawara is a Japanese open source enthusiast with a strong interest in desktop computing, particularly Ubuntu and LibreOffice. A former member of The Document Foundation, he also served as the Japanese translation lead for LibreOffice.\r\n\r\nBy day, he works as a security engineer, focusing on vulnerability assessment and DevSecOps. Within the open source community, he has contributed to various events and initiatives across Japan and Asia.\r\n\r\nHe has spoken at international conferences such as the FOSSASIA Summit, sharing his experiences as a non-native English speaker. He is particularly interested in making open source participation more accessible and sustainable through \"yuru\" (casual, low-pressure) engagement.",{"name":2892,"bio":2893},{"title":2896,"describe":2897,"type":30},"My broken English still works: speaking at global OSS events","I am not good at English, but I keep speaking at global OSS events.\r\nMy first international talk was a failure. I just read my script, and I could not follow the Q&A at all. Even after living in the United States for a year, my English did not improve much. But I learned something important: imperfect English can still work.\r\nAfter joining the LibreOffice community, I began speaking at international conferences. I often felt my English was not good enough. What changed me was a message from the community: sharing your perspective is valuable, even in broken English.\r\nIn this talk, I will share my journey, along with practical ideas you can use right away: keeping your English simple, reducing the text on your slides, focusing on one core message, and handling Q&A even when you don't understand everything.\r\nYou don't need perfect English to join the global OSS community. If you have something to share, your voice already has value.",{"title":2896,"describe":2897,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FUYUKW7",{"id":2902,"room":2903,"start":217,"end":499,"language":328,"track":2904,"speakers":2906,"zh":2928,"en":2931,"tags":2932,"uri":2933},"V8T9GX",{"en":1154,"zh-hant":10},{"id":330,"name":2905},{"en":332,"zh-hant":332},[2907,2914,2921],{"id":2908,"avatar":2909,"zh":2910,"en":2913},"ZL8GEL","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FZL8GEL_JDxiPlC.webp",{"name":2911,"bio":2912},"Grégory Schiano Lomoriello","Grégory is a seasoned technology professional with a passion for DevOps and innovations. With almost 20 years of experience working with software engineers and Infra\u002FOps engineers, I deeply understand the importance of collaboration and synergy between these teams. Currently, I'm a Senior Architect in the Platform engineering team at Canonical. In my role, I support the team in defining technical specifications, applying best practices, supporting team member carrier, and leading cross-team projects. Previously, as the CTO of Altissia, a company offering online language courses in 24 languages globally, I honed my leadership skills and developed a keen eye for technological solutions that enhance operational efficiency and user experience.",{"name":2911,"bio":2912},{"id":2915,"avatar":2916,"zh":2917,"en":2920},"NLA3YN","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FNLA3YN_OUUBc5u.webp",{"name":2918,"bio":2919},"Swetha Swaminathan","I started my career at Hewlett-Packard Enterprise working on control plane systems, where I developed an appreciation for the role strong platforms play in enabling cloud services at scale. Today, I’m a Software Engineer at Canonical, contributing to open-source projects by building Juju operators that power applications on Ubuntu.",{"name":2918,"bio":2919},{"id":2922,"avatar":2923,"zh":2924,"en":2927},"VNHNLR","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FVNHNLR_dpPnUt7.webp",{"name":2925,"bio":2926},"Javier de la Puente","I am a software engineer at Canonical working on software operators and infrastructure automation with Juju. My work centers on applying model-driven operations and 12-factor principles to simplify how diverse applications are built, configured, and operated at scale.",{"name":2925,"bio":2926},{"title":2929,"describe":2930,"type":59},"Deploy a 12-factor application of your choice end-to-end!","This workshop offers a hands-on exploration of the web applications lifecycle, demonstrating how to build and deploy cloud-native, 12-factor applications using a fully open-source toolchain. Participants can use their preferred framework (Go, Flask, ExpressJS, or Spring Boot) to gain practical experience taking an application from source code to a production-like environment.\r\n\r\nDuring the session, we will cover:\r\n - **Application Packaging:** Building standard OCI images for your\r\n   applications.\r\n - **Production Deployment:** Deploying workloads reliably across diverse\r\n   environments.\r\n - **Day-Two Operations:** Utilizing automated operators to manage complex\r\n   tasks like database backups and schema migrations.\r\n - **Infrastructure Flexibility:** Leveraging Rockcraft, Charmcraft, and\r\n   Juju to maintain platform independence, whether self-hosting or\r\n   utilizing a public cloud vendor.\r\n\r\nAdditionally, the workshop features a live demonstration of an AI Agent skill designed to automate these deployment steps, providing a simple, end-to-end approach to managing this open-source stack.\r\n\r\nMore info: https:\u002F\u002Fevents.canonical.com\u002Fevent\u002F146\u002Fcontributions\u002F933\u002F",{"title":2929,"describe":2930,"type":59},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FV8T9GX",{"id":2935,"room":2936,"start":731,"end":830,"language":96,"track":2937,"speakers":2939,"zh":2947,"en":2950,"tags":2951,"uri":2952},"VALWHR",{"en":684,"zh-hant":10},{"id":686,"name":2938},{"en":688,"zh-hant":689},[2940],{"id":2941,"avatar":2942,"zh":2943,"en":2946},"DULW37","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FDULW37_CQ0M7cb.webp",{"name":2944,"bio":2945},"李婉萍","政策與法律研究人員。",{"name":2944,"bio":2945},{"title":2948,"describe":2949,"type":30},"以CRA為例討論產品資安合規下的開源議題 (OSS Issue Map under CRA)","產品資安法規，正在重塑開源軟體生態的責任結構。歐盟《網路韌性法》（Cyber Resilience Act，CRA）之漏洞通報義務將於 2026 年 9 月正式生效，現有討論雖多聚焦於商業製造商的合規路徑，然開源人所面臨的系統性影響，實值等同關注。本討論以 CRA 為例切入，就資安合規框架對開源生態所觸發之議題進行梳理，期在 CRA 關鍵義務生效前，建構一份具參考意義的CRA「開源議題清單」（OSS Issue Map）。",{"title":2948,"describe":2949,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FVALWHR",{"id":2954,"room":2955,"start":453,"end":476,"language":96,"track":2956,"speakers":2958,"zh":2966,"en":2969,"tags":2972,"uri":2973},"VBEX7Y",{"en":67,"zh-hant":10},{"id":479,"name":2957},{"en":481,"zh-hant":481},[2959],{"id":2960,"avatar":2961,"zh":2962,"en":2965},"LJQFGG","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FLJQFGG_QkuaMFR.jpg",{"name":2963,"bio":2964},"王良丞\u002FLCWang","嵌入式系統工程師\u002Fembedded engineer",{"name":2963,"bio":2964},{"title":2967,"describe":2968,"type":30},"從 SPI 顯示面板到 Android 螢幕：ARM64 平台的 DRM 顯示驅動整合之旅","Android 的圖形顯示架構主要是針對 HDMI、eDP、MIPI-DSI 等高效能顯示介面所設計。然而，在許多嵌入式系統中，仍然會使用像 SPI 這類較為特殊的顯示介面，而這些介面通常並不在 Android 標準顯示管線的主要支援範圍內。\r\n\r\n本次演講將分享一個實際案例：如何在 ARM64 平台上，將一個基於 SPI 介面的自訂顯示驅動程式整合進 Android 14 的圖形顯示架構。該驅動程式是基於 Linux DRM（Direct Rendering Manager）框架實作，用於驅動一個透過 SPI 連接的小尺寸電子紙顯示器。\r\n\r\n演講將以 RK3588 平台為實際案例，說明 Linux DRM 子系統與 Android graphics stack 之間的整合流程。我們會解析 SurfaceFlinger、Hardware Composer（HWC3）與 DRM\u002FKMS 之間的互動關係，以及在 Android 系統中導入非標準顯示裝置時可能遇到的各種挑戰。",{"title":2970,"describe":2971,"type":30},"From SPI Panel to Android Screen: Integrating a Custom DRM Display Driver on ARM64","Android's graphics stack is primarily designed for high-performance display interfaces such as HDMI, eDP, and MIPI-DSI. However, many embedded systems still rely on unconventional display interfaces like SPI, which are rarely supported by Android's standard display pipeline.\r\n\r\nThis talk presents a real-world case study of integrating a custom SPI-based display driver into the Android 14 graphics stack on an ARM64 platform. The driver is implemented using the Linux DRM framework and drives a small e-paper display connected through SPI.\r\n\r\nUsing RK3588 as a practical example, this session walks through the full integration process between the Linux DRM subsystem and the Android graphics stack. We will explore how SurfaceFlinger, Hardware Composer (HWC3), and DRM\u002FKMS interact, as well as the challenges encountered when bringing up a non-standard display device in Android.",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FVBEX7Y",{"id":2975,"room":2976,"start":499,"end":500,"language":70,"track":2977,"speakers":2979,"zh":2987,"en":2990,"tags":2991,"uri":2992},"VFDKMV",{"en":1253,"zh-hant":10},{"id":901,"name":2978},{"en":903,"zh-hant":904},[2980],{"id":2981,"avatar":2982,"zh":2983,"en":2986},"BPBAUJ","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002Fbafkreidf6vutxmvxhzhwwfyjdymz3nhotfcmknoujvo7qglp3wkj43d3vu_B86Sz08.jpg",{"name":2984,"bio":2985},"Atsushi Eno","music software tools enthusiast. Audio Plugins for Android, Linux, MML, MIDI 1.0\u002F2.0, LV2. Kotlin Multiplatform. Music Tech Meetup Tokyo organizer.",{"name":2984,"bio":2985},{"title":2988,"describe":2989,"type":30},"design and implementation of a next-gen. DAW sequencer engine for desktop, mobile and web","Building a DAW or alike across multiple platforms including mobile and the Web has been a challenging task for various reasons. But thanks to recent innovation and development in audio dev. community, it became realistic in 2026.\r\n\r\nThrough this session we will discuss how cross-platform DAW sequencer engine became realistic, based on my own dev. experience. The topics will include:\r\n\r\n- inclusive audio plugin hosting API that covers Android and Web\r\n- audio engine requirements, especially node graphs and real-time audio processing boundary\r\n- GUI interaction approaches on each plugin format\r\n- how parameter updates work, especially across states and presets, among GUI, DSP and host\r\n- requirements on song project data models\r\n- separation of concerns among all these\r\n\r\nIn the agentic coding era, there are many emerging audio app projects that people used to find impossible to work on. I hope this session will give some guidance and design choices to those who want to build their own DAW or similar audio engines.",{"title":2988,"describe":2989,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FVFDKMV",{"id":2994,"room":2995,"start":2996,"end":2997,"language":96,"track":2998,"speakers":3000,"zh":3008,"en":3011,"tags":3014,"uri":3015},"VGHMGS",{"en":407,"zh-hant":10},"2026-08-08T10:05:00+08:00","2026-08-08T10:35:00+08:00",{"id":410,"name":2999},{"en":412,"zh-hant":412},[3001],{"id":3002,"avatar":3003,"zh":3004,"en":3007},"XMW9QB","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FXMW9QB_09Uhtvz.webp",{"name":3005,"bio":3006},"Uiiu","我叫 Uiiu，漢字寫作維佑。南投人。現在是軟體工程師，在新竹工作（身不由己）。專長是 Web 前端開發。開發之外的生活，對於語言學習、教育、嘻哈音樂感興趣。我會講台語、中文、英語、一點法語和排灣語。歡迎大家跟我交流。\r\n\r\nHi, I’m Uiiu. I’m from Nantou and currently based in Hsinchu, working as a Software Engineer specializing in Web Front-end Development.\r\n\r\nBeyond coding, I'm passionate about language learning, education, and Hip-hop. I speak Taiwanese, Mandarin, and English, plus a bit of French and Paiwan.\r\n\r\nFeel free to reach out—I’d love to connect!",{"name":3005,"bio":3006},{"title":3009,"describe":3010,"type":30},"實踐 Web 開發的技術平權：以 Local LLM 和 JS Runtime 打造人人可及的離線開發教室","我們習以為常的 AI 輔助開發或雲端生態系，對於身處矯正機構、偏鄉教育現場或數位資源受限環境的學習者而言，這道「連網門檻」可說是巨大的障礙。\r\n\r\n這場演講將展示一套基於 JavaScript 生態系構建的「邊緣開發環境」。在「完全斷網」的環境下，利用現有的開源工具，實現離線開發，且提供不輸給「網路吃到飽」的現代城市開發體驗。\r\n\r\n演講內容將深度解析如何整合以下關鍵技術組件，打造一套可單機運行的離線教室：\r\n\r\n1. **基礎設施去中心化**：利用 Verdaccio 建立本地端 Private npm Registry，預先快取並鏡像必要的依賴項，解決離線環境下無法 npm install 的焦慮。\r\n\r\n2. **知識庫 RAG 化**：將 MDN (Mozilla Developer Network) 靜態文件向量化，透過 RAG (Retrieval-Augmented Generation) 技術，在沒有 Google 與雲端 AI 的情況下，提供精準的離線技術檢索。\r\n\r\n3. **AI 導師在地化**：透過 Ollama 在本地端部署輕量化 LLM (如 Llama 3 或 DeepSeek-Coder)，並整合 Aider 與 Node.js Runtime，實踐 AI 協作開發、自動化修復 Bug 的現代工作流。\r\n\r\n希望透過本分享傳達一個核心理念：這些看似尖端的技術並非遙不可及，透過合理的系統架構與整合，我們能將技術的火種帶進任何有心學習的角落，真正實現「技術平權」。",{"title":3012,"describe":3013,"type":30},"Web Tech Equality: Building an Offline Web Dev Classroom with Local LLM and JS Runtime","High-speed internet and AI-powered cloud tools are now standard for developers. However, for learners in correctional facilities, rural areas, or resource-limited environments, this \"connectivity barrier\" remains a major obstacle.\r\n\r\nThis session demonstrates an \"Edge Development Environment\" built on the JavaScript ecosystem. We will show how to use existing open-source tools to create a modern development experience that rivals high-speed city environments—even in total offline conditions.\r\n\r\nWe will deep dive into the integration of these three core components:\r\n\r\nDecentralized Infrastructure: Using Verdaccio to build a local Private npm Registry, solving the \"npm install\" issue without an internet connection.\r\n\r\nRAG-based Knowledge Base: Vectorizing MDN documentation with RAG (Retrieval-Augmented Generation) to provide precise technical search without Google or Cloud AI.\r\n\r\nLocal AI Mentor: Deploying local LLMs via Ollama and integrating with Aider, enabling a modern workflow for AI pair programming and automated bug fixing.\r\n\r\nOur core mission: Advanced technology should not be a privilege. Through smart system architecture, we can bring the spark of technology to any corner, achieving true \"Tech Equality.\"",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FVGHMGS",{"id":3017,"room":3018,"start":217,"end":1278,"language":328,"track":3019,"speakers":3021,"zh":3029,"en":3032,"tags":3033,"uri":3034},"VJFBML",{"en":326,"zh-hant":10},{"id":330,"name":3020},{"en":332,"zh-hant":332},[3022],{"id":3023,"avatar":3024,"zh":3025,"en":3028},"JB8TBF","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FJB8TBF_dAW4Td6.webp",{"name":3026,"bio":3027},"Shishir Subedi","Shishir is a security engineer at Canonical, where he works on mitigating vulnerabilities across applications in the Ubuntu archive. He holds a degree in computer engineering and has hands-on experience in open-source security, application hardening, and infrastructure security. When not at a terminal, he can be found on a football pitch, at a table tennis table, or on a hiking trail.",{"name":3026,"bio":3027},{"title":3030,"describe":3031,"type":30},"Beyond apt install: The Journey of an Ubuntu Package","Every day, millions of users type `apt update` and `apt install` and software appears on their system as if by magic. Behind this simple command lies a sophisticated infrastructure of cryptographic safeguards, precise archive organization, and rigorous human review. This talk is an invitation to look under the hood and trace the full lifecycle of an Ubuntu package, from the central archive to the local machine.\r\n\r\nThe journey begins with an exploration of the repository architecture. We will break down how your system's repository configuration maps to a structured archive where metadata and installable packages are organized separately. We will discuss how Ubuntu manages change over a release's lifetime through distinct channels for security fixes, stable updates, and backported features, and how the archive groups packages by the level of support and maintenance they receive.\r\n\r\nTrust is the backbone of this ecosystem. We will demonstrate the \"Chain of Trust\" that ensures package authenticity, illustrating how the system verifies that software is genuine and untampered with, even when fetched over potentially untrusted network mirrors. This section illustrates how a single signed file at the top anchors the trust for every package in the archive.\r\n\r\nFinally, we will explore the human \"gatekeepers\" of the ecosystem. We will demystify the procedural checks and balances, such as the security auditing and stable release update processes that ensure software remains reliable long after a version is released.\r\n\r\nMore info: https:\u002F\u002Fevents.canonical.com\u002Fevent\u002F146\u002Fcontributions\u002F946\u002F",{"title":3030,"describe":3031,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FVJFBML",{"id":3036,"room":3037,"start":12,"end":382,"language":70,"track":3038,"speakers":3040,"zh":3048,"en":3051,"tags":3052,"uri":3053},"VL8BGL",{"en":121,"zh-hant":10},{"id":733,"name":3039},{"en":735,"zh-hant":735},[3041],{"id":3042,"avatar":3043,"zh":3044,"en":3047},"UUFM8S","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FUUFM8S_MfF0mtb.webp",{"name":3045,"bio":3046},"Lucía Luzuriaga","Lucía is a Engineering and Technical Program Manager. \r\nWith vast experience in thr software engineering field and an honorable degree in Economics, she bridges the gap between software teams and non-technical partners by representing developers, removing blockers, and translating technical challenges into strategic initiatives that support organizational impact.",{"name":3045,"bio":3046},{"title":3049,"describe":3050,"type":30},"Open source, open voice: the power of FOSS in supporting citizen participation","How can open source strengthening democracy in practice? \r\nThis talk shares insights from CONSUL - a FOSS platform for citizen participation - in how instance implementations can make impact in supporting institutions and citizens worldwide, as well as the challenges of resource constraints, strategic prioritization and sustaining FOSS communities. We’ll also touch on how AI can enhance civic processes while remaining transparent and accountable.",{"title":3049,"describe":3050,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FVL8BGL",{"id":3055,"room":3056,"start":95,"end":1155,"language":328,"track":3057,"speakers":3059,"zh":3074,"en":3077,"tags":3078,"uri":3079},"VREQUY",{"en":326,"zh-hant":10},{"id":330,"name":3058},{"en":332,"zh-hant":332},[3060,3067],{"id":3061,"avatar":3062,"zh":3063,"en":3066},"JHZFQ9","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FJHZFQ9_J57cOq2.webp",{"name":3064,"bio":3065},"Ronit Banerjee","Ronit Banerjee is a DevOps Engineer (R&D) at Keysight Technologies and a CNCF Ambassador passionate about cloud-native ecosystems and open source. Based in Kolkata, India, he leads the Cloud Native Hooghly community chapter. Ronit has been a Google Summer of Code Mentor at DBpedia since 2024. He is also a UbuCon Global Committee member and organiser, and has volunteered with CNCF's SIG Contribex at KubeCon Contributor Summits.",{"name":3064,"bio":3065},{"id":3068,"avatar":3069,"zh":3070,"en":3073},"GZ7ADG","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FGZ7ADG_AO3PW4H.webp",{"name":3071,"bio":3072},"Rishi Mondal","I'm an SRE at Obmondo, CNCF KubeStellar Maintainer, and Docker Captain. A two-time Google Summer of Code contributor and Research Intern at UC Santa Cruz. As a Linux Foundation Mentor, GSoC Mentor at CNCF, and OWASP Mentor, I uniquely progressed from LFX Mentee to Maintainer to Mentor within a single year. A regular cloud-native speaker at multiple events across Asia, I was personally featured by the CNCF CEO at KubeCon India for community impact. Harvard HPAIR Delegate and LIFT Scholar.",{"name":3071,"bio":3072},{"title":3075,"describe":3076,"type":30},"Kubernetes Is Not a Platform: Building One with Juju","Installing Kubernetes is easy. Operating it across teams, under pressure, at scale that's where most organizations quietly struggle.\r\n\r\nClusters grow into a mix of manual `kubectl` workflows, inconsistent network policy, and operational knowledge nobody wrote down. It works until it doesn't.\r\n\r\nThe missing layer is **operational lifecycle management** and that's exactly what **Juju** was built for. **Juju** brings model-driven operations to Kubernetes: it manages service dependencies, handles upgrades and integrations, automates day-2 recovery, and encodes operational knowledge into reusable charms rather than fragile scripts and runbooks. Where most tools stop at deployment, **Juju** carries the operational story forward through the entire application lifecycle.\r\n\r\nBut a platform needs more than lifecycle automation. Delivery needs to be declarative and auditable that's where **Argo CD** fits in, providing GitOps-based deployment and drift control. And networking and security policy need to be enforceable at runtime that's where **Cilium** brings kernel-level policy enforcement and observability.\r\n\r\nThis session shows how **Juju** serves as the operational backbone of a Kubernetes platform, with Argo CD and Cilium filling the delivery and networking layers. Rather than presenting tools in isolation, the talk demonstrates a cohesive platform engineering model built around Juju's strengths in lifecycle automation and operational relationships.\r\n\r\nAttendees will see how **Juju's** charm-based approach solves problems that GitOps alone cannot service integrations, cross-application dependencies, upgrade orchestration, and recovery workflows while Argo CD and Cilium handle what they do best.",{"title":3075,"describe":3076,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FVREQUY",{"id":3081,"room":3082,"start":830,"end":1132,"language":96,"track":3083,"speakers":3085,"zh":3093,"en":3096,"tags":3097,"uri":3098},"W39FHQ",{"en":922,"zh-hant":10},{"id":924,"name":3084},{"en":926,"zh-hant":927},[3086],{"id":3087,"avatar":3088,"zh":3089,"en":3092},"GK99EM","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FGK99EM_skiQSvb.jpg",{"name":3090,"bio":3091},"Ronny Wang","g0v 長期參與者，喜歡爬政府資料並將政府資料變成好用乾淨的資料，曾做過台灣公司資料、開放政治獻金、新聞小幫手等專案，2023 年成立歐噴有限公司，想透過公司來更永續的推動開源文化，並且讓政府更透明，有開發 LawTrace 和 OpenBudget 等專案讓法律歷程及政府預算更好取用",{"name":3090,"bio":3091},{"title":3094,"describe":3095,"type":30},"開公司推廣開源，有搞頭嗎？","g0v 長期參與者 Ronny 在 2023 年成立了歐噴有限公司（OpenFun），期待將過去在 g0v 的興趣，轉變為公司的目標，讓開放政府運動能更永續的進行下去。\r\n\r\n三年來歐噴逐漸站穩腳步，開發了 LawTrace 和 OpenBudget 等開放政府開源專案，並且持續在與政府倡議開放協作文化，歐噴是怎麼走過來的呢？",{"title":3094,"describe":3095,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FW39FHQ",{"id":3100,"room":3101,"start":122,"end":238,"language":328,"track":3102,"speakers":3104,"zh":3108,"en":3111,"tags":3112,"uri":3113},"W3RQUV",{"en":922,"zh-hant":10},{"id":946,"name":3103},{"en":948,"zh-hant":949},[3105],{"id":2655,"avatar":2656,"zh":3106,"en":3107},{"name":2658,"bio":2659},{"name":2658,"bio":2659},{"title":3109,"describe":3110,"type":30},"How MySQL Is Evolving: Open Roadmaps, Community Collaboration, and New Ways to Contribute","MySQL is evolving its development process and community engagement. Oracle is shifting from a model where it primarily defined the roadmap internally and released the results as open source, toward a more open, community-oriented approach.\r\n\r\nThis session introduces MySQL’s Community Engagement Strategy and Community Roadmap, explaining how Oracle is working to expand the MySQL ecosystem by making roadmap discussions more transparent and encouraging broader participation from users, developers, DBAs, and contributors.\r\n\r\nWe will cover how the MySQL community can engage with the project through public discussions, feedback on bugs and feature requests. Attendees will learn how they can help shape the future of MySQL—not only as users of the database, but as active participants in its ongoing evolution.",{"title":3109,"describe":3110,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FW3RQUV",{"id":3115,"room":3116,"start":216,"end":217,"language":96,"track":3117,"speakers":3119,"zh":3127,"en":3130,"tags":3131,"uri":3132},"WBWFTB",{"en":192,"zh-hant":10},{"id":196,"name":3118},{"en":198,"zh-hant":198},[3120],{"id":3121,"avatar":3122,"zh":3123,"en":3126},"83A7DR","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002F83A7DR_N2Jcfeg.webp",{"name":3124,"bio":3125},"Yu-Chien Peter Lin","Peter Lin works on system software, firmware, and security — with a focus on the low-level engineering that makes trustworthy computing platforms actually work. His current work centers on hardware isolation and secure execution environments in the open RISC-V ecosystem.",{"name":3124,"bio":3125},{"title":3128,"describe":3129,"type":30},"Architecting Trusted Execution on RISC-V: Worlds, WorldGuard and Supervisor Domains","As system-level hardware isolation has long been a missing piece in the RISC-V ecosystem, emerging standards are now closing that gap. This session provides a general introduction to deploying hardware-enforced Trusted Execution Environments (TEEs) using the RISC-V Worlds ISA extension and the WorldGuard architecture. We will explore how initiators tag execution environments and how target resources enforce access control rules. Transitioning to software, we will cover defining isolation boundaries via Device Tree bindings and managing system domains within OpenSBI. Finally, we will demonstrate integrating this security stack with OP-TEE, leveraging the RPMI TEE Service Group to establish standardized cross-world communication.",{"title":3128,"describe":3129,"type":30},[402],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FWBWFTB",{"id":3134,"room":3135,"start":1112,"end":454,"language":42,"track":3136,"speakers":3138,"zh":3146,"en":3149,"tags":3150,"uri":3151},"WK8EEE",{"en":39,"zh-hant":10},{"id":384,"name":3137},{"en":386,"zh-hant":10},[3139],{"id":3140,"avatar":3141,"zh":3142,"en":3145},"ZJ98MX","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FZJ98MX_woQ77ap.webp",{"name":3143,"bio":3144},"Jennifer HSU","目前任職於自託管隱私錢包 Fluidkey，也是台灣開發者社群 XueDAO 的創辦人，致力於推廣區塊鏈技術，讓更多人能一起探索與討論這個領域。",{"name":3143,"bio":3144},{"title":3147,"describe":3148,"type":30},"從不可連結性出發：隱匿地址如何解決鏈上金融隱私 — 以 Fluidkey 為例","區塊鏈的透明性是雙面刃：任何人都能查詢一個地址的完整交易歷史與餘額。對於日常支付、薪資發放或商業往來，這種「永久公開的財務紀錄」並不符合現實世界對隱私的基本需求。\r\n\r\n隱匿地址（Stealth Address） 是一種密碼學方案。收款方公開一組由 spending public key 與 viewing public key 組成的 stealth meta-address。每次需要新的收款地址時，會先產生一個 ephemeral key，再透過 ECDH 與收款方的 viewing public key 算出 shared secret，最後結合 spending public key 推導出一個全新的一次性 stealth EOA 作為實際收款地址。其結果是：\r\n- 即使同一個收款人收到一百筆款項，這些款項會分別落在 一百個彼此無關的鏈上地址 中，外部觀察者無法將它們連結到同一個身份\r\n- 只有持有對應 viewing key 的收款人，能掃描鏈上交易並識別出「哪些地址屬於我」，再用 spending key 動用資金\r\n\r\n這段分享中，我會從 ECDH 金鑰交換出發，拆解 ERC-5564 的隱匿地址推導流程，並說明為什麼隱匿地址提供的是 **「不可連結性（unlinkability）」而非「不可追溯性（untraceability）」**。\r\n這個區分很關鍵：每一筆交易在鏈上依然完整可見、金額與資金流向都可追溯、可審計；被切斷的只是「地址與收款人身份」、以及「同一收款人的不同地址之間」的關聯。這讓隱匿地址在需要選擇性揭露的場景（例如稅務申報、審計）下，比全匿名方案更具實用性 — 使用者保有對自身資料揭露的主動權。\r\n\r\n分享時我會以 Fluidkey 為例。Fluidkey 在 ERC-5564 的基礎上做了一項關鍵設計：將 ephemeral key 改為從 viewing key node 以 BIP-32 階層式路徑 deterministic 推導而來，而非隨機產生。這個看似微小的工程選擇，帶來一個重要的安全性保證 — 使用者可以完全獨立地重放所有歷史 stealth address 並恢復資金，不需要依賴 Fluidkey 或任何中心化服務。這讓隱匿地址不只是隱私工具，而是真正符合 self-custody 精神的隱私基礎設施。",{"title":3147,"describe":3148,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FWK8EEE",{"id":3153,"room":3154,"start":499,"end":500,"language":70,"track":3155,"speakers":3157,"zh":3165,"en":3168,"tags":3169,"uri":3170},"WMPHPN",{"en":522,"zh-hant":10},{"id":524,"name":3156},{"en":526,"zh-hant":10},[3158],{"id":3159,"avatar":3160,"zh":3161,"en":3164},"PB8MZF","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FPB8MZF_rAU6Tj3.png",{"name":3162,"bio":3163},"Marvin Drees","BMC and RoT firmware developer at 9elements, eager to learn and teach new technologies.",{"name":3162,"bio":3163},{"title":3166,"describe":3167,"type":30},"Tamago capabilities and why you should run Go-based firmware","This talk will showcase and explore the usage of Go outside the normal usage in an operating system and how you can even write bare metal firmware with it. We will show how to run VPN tools like Tailscale and Netbird directly on your hardware without an operating system. We will also learn what has changed since the last Tamago talk during OSFC last year including new platforms, upstream effort of bare metal Go directly into the official compiler and pure-go networking and filesystem work.",{"title":3166,"describe":3167,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FWMPHPN",{"id":3172,"room":3173,"start":122,"end":3174,"language":96,"track":3175,"speakers":3177,"zh":3178,"en":3181,"tags":3182,"uri":3183},"WPRG3M",{"en":1253,"zh-hant":10},"2026-08-08T10:15:00+08:00",{"id":1258,"name":3176},{"en":1260,"zh-hant":1261},[],{"title":3179,"describe":3180,"type":30},"[Prime Session] Day1 Morning Session","Absctract",{"title":3179,"describe":3180,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FWPRG3M",{"id":3185,"room":3186,"start":408,"end":327,"language":96,"track":3187,"speakers":3189,"zh":3204,"en":3207,"tags":3208,"uri":3209},"WRT7WL",{"en":9,"zh-hant":10},{"id":15,"name":3188},{"en":17,"zh-hant":18},[3190,3197],{"id":3191,"avatar":3192,"zh":3193,"en":3196},"7QZY9Y","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002F7QZY9Y_cJxdGMk.webp",{"name":3194,"bio":3195},"Riley","我會在稍後補上，造成不便敬請見諒。",{"name":3194,"bio":3195},{"id":3198,"avatar":3199,"zh":3200,"en":3203},"PWEMYR","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FPWEMYR_Lw0ntSa.webp",{"name":3201,"bio":3202},"Bradly","TBD",{"name":3201,"bio":3202},{"title":3205,"describe":3206,"type":30},"用社群的力量打造服務所有人的 App：翻譯平台整合撞牆談","我和北科程式設計研究社的伙伴利用課餘（以及偶爾的上課時間）維護一款 GPLv3 授權的校園軟體 Tattoo，協助北科大學生查課表、看成績、處理校務系統。專案從 zh-TW 與 en-US 雙語起步，英文由社團成員人工翻譯；後來為了服務校內多數來自非英語系國家的外籍生，又逐步擴充到我們自己不熟的語言,採用「LLM 出初稿、再透過翻譯平臺交給該語系學生修正」的流程。半年下來繞了不少遠路，這場分享想把走過的路攤開，給做小型自由軟體本地化的開發者一份參考。\r\n\r\n一開始談**翻譯協作平臺**。為了上述的多語流程，我先後試過 Weblate 和 Crowdin，最後兩個都關掉了。一方面是同步機制本身彆扭：譯者在平臺上的修改會被 bot 一筆筆推回 GitHub，被「New translations zh-tw.i18n.yaml」這類 commit 淹沒主線。另一方面是兩個平臺接 GitHub 的權限模型：為了一個專案的翻譯，卻得授權整個組織的管理權限，這個取捨很難說服自己。最後我們只能暫時關閉同步，並重新思考要怎麼讓非開發者的譯者貢獻。\r\n\r\n我們還會介紹 **Flutter 上的選擇**。Flutter 生態有幾套常見的 i18n 套件，我們最後選了對譯者與審查者格式友善的 slang。框架本身不是這場的重點，只會用幾分鐘帶過技術選型的過程。\r\n\r\n接下來我們會討論**翻譯檔的儲存格式**。從 ARB (JSON-based) 看到 JSON、YAML，討論為什麼落腳在 YAML：人類可讀、可註解、多行字串友善、不被尾逗號搞死。當然 YAML 並非完美方案：行尾多餘空格、長字串折行語意、保留字與特殊字元的 escaping 都會偶爾造成問題，這些經驗一併分享。\r\n\r\n最後是**漢英對譯實戰陷阱**：以漢語為母語的開發者很容易忘了其他語系有單複數、時態等語法分支——當你直覺地把字串寫好後，等到 en-US 要區分 `one`\u002F`other` 時，就得回頭重寫整個結構，整個 App 一起改。我會以「密碼將於 N 天後過期」為例，示範從第一天就該怎麼設計翻譯鍵。同場加映 CJK 與英數字之間的空格該不該寫進原始字串、教室名稱與專有名詞的雙語對應策略，以及校務系統回傳混雜中英資料時，如何用一個小小的 `localized(zh, en)` helper 處理。\r\n\r\n希望這 30 分鐘能讓做 Flutter App 或其他軟體的朋友，少踩幾個我們踩過的坑。",{"title":3205,"describe":3206,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FWRT7WL",{"id":3211,"room":3212,"start":95,"end":1155,"language":328,"track":3213,"speakers":3215,"zh":3223,"en":3226,"tags":3227,"uri":3228},"WXCPLL",{"en":1007,"zh-hant":10},{"id":901,"name":3214},{"en":903,"zh-hant":904},[3216],{"id":3217,"avatar":3218,"zh":3219,"en":3222},"XL9XGX","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FXL9XGX_MKU9lqf.webp",{"name":3220,"bio":3221},"Anant Awasthi","I am a DevSecOps engineer with hands-on experience in Elasticsearch, Docker, and Kubernetes. I work on building secure and scalable systems, focusing on observability, logging, and monitoring solutions. I have practical experience in deploying and managing Elastic Stack in real-world environments and enjoy sharing knowledge with the tech community.",{"name":3220,"bio":3221},{"title":3224,"describe":3225,"type":30},"Observability on Ubuntu with Elastic Stack","Observability plays a key role in understanding and maintaining modern systems. In this session, I will demonstrate how to build a complete observability setup on Ubuntu using the Elastic Stack.\r\n\r\nThe talk will cover how to collect and analyze logs, system metrics, and application data, and visualize them using Kibana dashboards. I will also share practical insights from real-world implementations, including common challenges, performance considerations, and best practices.\r\n\r\nAttendees will learn how Ubuntu provides a stable and efficient foundation for running observability workloads, and how to use open-source tools to monitor, troubleshoot, and improve system reliability.",{"title":3224,"describe":3225,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FWXCPLL",{"id":3230,"room":3231,"start":499,"end":500,"language":96,"track":3232,"speakers":3234,"zh":3238,"en":3241,"tags":3242,"uri":3243},"WYGFAZ",{"en":263,"zh-hant":10},{"id":44,"name":3233},{"en":46,"zh-hant":47},[3235],{"id":268,"avatar":269,"zh":3236,"en":3237},{"name":47,"bio":271},{"name":47,"bio":271},{"title":3239,"describe":3240,"type":30},"為什麼匿名支付重要：開源、去中心化的金流與台灣 VASP 法 2026","談匿名與隱私時，金流常被遺漏。一筆轉帳的時間、金額、收款對象，配上信用卡號或銀行戶名，幾乎能還原一個人的社交網絡與活動範圍。這場說明金流為什麼是最難擺脫的一種 metadata：開戶需要實名 KYC（金融機構確認客戶身分的程序）、紀錄依法長期保存、跨機構雙向留存、在合法程序下可被調閱。也會談金流如何被當成審查工具，從 PayPal、Venmo 凍結特定帳號的案例，到長年架設 Tor 中繼節點、卻被支付平台無預警關停的處境，凸顯把支付完全託付給封閉的商業金融中介有什麼風險。對照之下，這場盤點以開源協定為基礎的替代方案，例如協議層就遮蔽交易資訊的隱私幣 Monero、Zcash，以及零知識證明（zero-knowledge proof，不透露內容本身就能證明某件事為真的密碼學方法）、多重簽署等開源實作，並逐一點出各自在技術門檻、合法性與接受度上的取捨。最後結合台灣 2026 年上路的虛擬資產服務法（VASP 法）說明在地脈絡。社群一向強調合法是匿名支付的前提，工具用於教育與理解風險。這場為下午與 ETHTaipei 合辦的匿名支付場暖身。",{"title":3239,"describe":3240,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FWYGFAZ",{"id":3245,"room":3246,"start":2174,"end":3247,"language":96,"track":3248,"speakers":3250,"zh":3258,"en":3261,"tags":3262,"uri":3263},"WZLMDE",{"en":922,"zh-hant":10},"2026-08-08T14:35:00+08:00",{"id":924,"name":3249},{"en":926,"zh-hant":927},[3251],{"id":3252,"avatar":3253,"zh":3254,"en":3257},"9H8YD8","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002F9H8YD8_apCGI9Z.webp",{"name":3255,"bio":3256},"Jimmy Huang","使用開源軟體服務NPO將近20年，2009年開始創辦網絡行動科技（股）有限公司，為共同創辦人。",{"name":3255,"bio":3256},{"title":3259,"describe":3260,"type":30},"對我們的客戶來說，SaaS 開源重要嗎？","網絡行動科技服務NPO領域多年，從公司開始至今都是使用開源軟體（授權以AGPL為主）架設網站，提供服務，若客戶擁有技術人員，服務也可供技術人員自行架設應用。\r\n\r\n但對我們的客戶來說，這件事重要嗎？本次想聊聊開源文化，如何深深影響我們的商業模式。\r\n\r\n在公司內部，無論從軟體開發到工作決策上，有了開源（軟體、文化），會有什麼不一樣。\r\n也想聊聊以我們的屬性來說，開源軟體在 SaaS 商業營運時，客戶在意的內容，跟開源有多大差距。\r\n最後可以講我們近幾年的嘗試，開源對 NPO 領域，我們或有不同的切入點和想法。",{"title":3259,"describe":3260,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FWZLMDE",{"id":3265,"room":3266,"start":94,"end":41,"language":42,"track":3267,"speakers":3272,"zh":3280,"en":3283,"tags":3284,"uri":3285},"X7P83A",{"en":167,"zh-hant":10},{"id":3268,"name":3269},543,{"en":3270,"zh-hant":3271},"Open LLM End User: Downstream Applications","Open LLM End User: 開源模型應用",[3273],{"id":3274,"avatar":3275,"zh":3276,"en":3279},"WJJQHU","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FWJJQHU_cwnzFj1.webp",{"name":3277,"bio":3278},"Tan Lean Sheng","Analog player in the digital world. Technologist and AI evangelist for greater good. Open source community contributor and maintainer for almost a decade.",{"name":3277,"bio":3278},{"title":3281,"describe":3282,"type":59},"迷你 AI 黑客松 (Mini AI Hackathon)：打造一個能做一件實事的 Agent","大部分 AI agent 的 demo 在台上很厲害，一碰到真實任務就沒用了。這場 2 小時的迷你 hackathon 要做的正好相反：一起做一個小小的、但真的能動的東西。\r\n我們會給一個明確的題目和一份起始素材（starter kit），讓你不用從零開始——現場跟著一起打造一個能從頭到尾完成任務的 AI agent。範圍抓小、把它做出來、現場 demo。\r\n不需要事前準備、也不需要 agent 經驗，帶著筆電和好奇心來就好。歡迎使用開源權重模型與開源工具。可以單人做，也可以現場組隊，會有 mentor 在場協助。\r\n流程上，我們會先說明題目與規則，大部分時間拿來動手做，最後以簡短的現場 demo 收尾。來做一個東西、認識其他開發者，看看一整個房間的人在兩小時內能生出什麼。\r\n關於題目細節、建議工具與 starter kit，我們會在活動前另行公布。",{"title":3281,"describe":3282,"type":59},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FX7P83A",{"id":3287,"room":3288,"start":709,"end":68,"language":96,"track":3289,"speakers":3291,"zh":3299,"en":3302,"tags":3303,"uri":3304},"X9SGLP",{"en":899,"zh-hant":10},{"id":901,"name":3290},{"en":903,"zh-hant":904},[3292],{"id":3293,"avatar":3294,"zh":3295,"en":3298},"NR8XWJ","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FDong_pnbA3vm.jpg",{"name":3296,"bio":3297},"Dong","在開源社群第三年的開源新鮮人。\r\n\r\n今年在 g0v Summit 2026 的坑裡，\r\n參與宣傳、設計的傳接球訓練。\r\n\r\n最大的希望，\r\n是丟出去的球都有人接，\r\n也希望自己能好好對待每顆丟來的球。",{"name":3296,"bio":3297},{"title":3300,"describe":3301,"type":30},"當設計思維遇上開源營運：讓志工更優雅協作的小工具實踐","在開源社群與大型年會（如 g0v Summit）的籌備過程中，往往會面臨到「志願者能量有限」與「行政庶務繁瑣」的雙重挑戰。當我們用設計思維重新審視這些痛點時，會發現許多摩擦力來自於格式轉譯與重複勞動。\r\n\r\n本議程將分享如何透過「服務設計」的視角，開發一系列輕量化工具：從可以驗證信箱地址的自動化折扣碼發放系統、讓編輯者能在雲端文件工具排版並一鍵轉為 HTML 郵件寄出的小工具，到將結構化資料（Opass json）自動生成為 SVG 檔的互動式網頁（[g0v\u002Fopass-schedule-to-svg](https:\u002F\u002Fgithub.com\u002Fg0v\u002Fopass-schedule-to-svg)）。\r\n\r\n我們將探討如何透過這些「看不見的設計」，降低參與門檻，讓志工能將珍貴的創造力留在核心任務上，實現真正「優雅」的開源協作。",{"title":3300,"describe":3301,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FX9SGLP",{"id":3306,"room":3307,"start":382,"end":453,"language":96,"track":3308,"speakers":3310,"zh":3317,"en":3320,"tags":3323,"uri":3324},"XFJBFU",{"en":350,"zh-hant":10},{"id":352,"name":3309},{"en":354,"zh-hant":355},[3311],{"id":3312,"avatar":3313,"zh":3314,"en":3316},"WR9SYU","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FF1070001_sTYmUqj.jpg",{"name":3315,"bio":10},"劉維人",{"name":3315,"bio":10},{"title":3318,"describe":3319,"type":30},"架構即憲法：以法定規格與可分叉性，確保身分驗證系統保障數位權利","身分是行使權利的基礎，而在數位系統中，程式碼的能力定義了權利與權力的邊界。當底層架構讓個人資料的跨脈絡串接唾手可得，隱私與秘密就不再受到民主法治的保護。為了保障相應權利、守護數位主權，必須從**法定規格 (Statutory Spec)** 出發，將民主價值轉化為系統強制力：\r\n\r\n1. **身分驗證不是數位監控**：以「控制權驗證」取代傳統的「屬性覆核」，確保使用者負責，同時保留公民「拒絕被連結」的權利。\r\n2. **技術可行即規範義務**：在身分驗證中，若技術上能做到不蒐集資料，法律上就「應該」禁止蒐集。隱私強化技術 (PETs) 必須從「最佳實踐」提升為「法定義務」。\r\n3. **數位主權的終極防線**： **「可分叉性 (Forkability)」** 與 **「可重建性 (Rebuildability)」** 必須成為抵抗技術鎖定的核心。建立對生產環境的實質審查機制，確保國家與社群擁有獨立運作系統的能力。",{"title":3321,"describe":3322,"type":30},"Architecture as Constitution: Safeguarding Digital Rights via Statutory Specs and Forkability","Identity is the foundation for exercising rights. In digital systems, the capabilities of code define the boundaries of both rights and power. When underlying infrastructure makes cross-context data correlation trivial, privacy and secrecy are no longer protected by the rule of law. To safeguard these rights and defend Digital Sovereignty, we must start with Statutory Specifications, translating democratic values into enforceable system constraints.\r\n\r\nThis session explores how to embed democratic principles into system architecture through Statutory Specs:\r\n\r\nIdentity Verification ≠ Digital Surveillance:\r\nWe propose shifting from traditional \"Attribute Verification\" to \"Control Verification.\" This paradigm ensures accountability while strictly preserving the citizen's right to unlinkability (the right to refuse to be linked).\r\n\r\nTechnical Feasibility Implies Normative Obligation:\r\nIf it is technically feasible to verify identity without data collection, the law must prohibit such collection. We argue that Privacy-Enhancing Technologies (PETs) must evolve from optional \"Best Practices\" into mandatory \"Statutory Obligations.\"\r\n\r\nThe Ultimate Line of Defense:\r\nForkability and Rebuildability are not just open-source concepts but the core of resisting technical lock-in. We must establish substantive auditing mechanisms for production environments, ensuring that the state and the community retain the independent capability to operate and reconstruct their systems.",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FXFJBFU",{"id":3326,"room":3327,"start":216,"end":217,"language":42,"track":3328,"speakers":3330,"zh":3338,"en":3341,"tags":3342,"uri":3343},"XPXN9A",{"en":67,"zh-hant":10},{"id":72,"name":3329},{"en":74,"zh-hant":74},[3331],{"id":3332,"avatar":3333,"zh":3334,"en":3337},"RR3SMW","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FRR3SMW_kHqEhx6.png",{"name":3335,"bio":3336},"Homun Mage","Senior AI Engineer (Infotech). Loving Svelte and FastAPI.\r\n\r\n▶ 🐱 github: [homunmage](https:\u002F\u002Fgithub.com\u002Fhomunmage)\r\n▶ 📧 mail: homunmage@gmail.com\r\n▶ 🌍 telegram: posetmage\r\n▶ 💬 line: posetmage",{"name":3335,"bio":3336},{"title":3339,"describe":3340,"type":30},"工程師最陌生的熟人 - 文字二三事","雖然我們每天都在打code，但是文字一直以來都是大坑，分享各種我與各種文字相關場景打交道的心得\r\n* 玩日文遊戲需要中文字幕 -> OCR\r\n* 看網頁文章想要有聲音 -> TTS\r\n* 想要字幕 -> STT 語音轉文字 whisper\r\n* 寫部落格、筆記系統\r\n* pdf跑版 -> 輸入內容沒版型 -> latex or html\r\n* 想用markdown寫一本書? -> pipeline",{"title":3339,"describe":3340,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FXPXN9A",{"id":3345,"room":3346,"start":12,"end":382,"language":70,"track":3347,"speakers":3349,"zh":3357,"en":3360,"tags":3362,"uri":3363},"XRJJV3",{"en":1253,"zh-hant":10},{"id":901,"name":3348},{"en":903,"zh-hant":904},[3350],{"id":3351,"avatar":3352,"zh":3353,"en":3356},"G8M9GY","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FG8M9GY_9sZt88r.webp",{"name":3354,"bio":3355},"Yusuke Ohmori","I am a software engineer and community organizer from Japan, currently based abroad. I run “Engineer Anime（エンジニアニメ）,” a community where engineers share lessons they have gained from anime and connect them to professional practice.",{"name":3354,"bio":3355},{"title":3358,"describe":3359,"type":30},"Designing a Cross-Disciplinary Engineering Community Through Anime","I run **“Engineer Anime,”** a community where engineers share lessons they have learned from anime. In this community, we organize events where engineers give presentations about insights they gained from anime and connect them to their own experiences in software development and engineering work.\r\n\r\nAt these events, speakers present ideas inspired by anime stories, characters, or production processes, and relate them to topics such as software development, organizational design, and career growth. For example, some speakers discuss **teamwork and organizational structure** by examining how teams collaborate in anime. Others reflect on **engineering careers and learning** through the personal growth and struggles of characters. There are also talks that use the **anime production process** as a starting point to discuss project management and team structures.\r\n\r\nUsing anime as a shared cultural reference helps lower the barrier to discussion. Participants with different backgrounds and experience levels—from students and early-career engineers to experienced professionals—can easily join the conversation. Because many people are already familiar with anime stories, discussions often start from something relatable and naturally expand into deeper reflections about engineering practices, teamwork, and personal growth.\r\n\r\nThis approach also encourages **cross-disciplinary dialogue**. In addition to technical topics, discussions often explore broader themes such as craftsmanship, collaboration, leadership, and how people grow through challenges—topics that are not always emphasized in traditional technical events.\r\n\r\nIn this session, I will introduce how a community built around lessons from anime can create meaningful conversations among engineers. Through concrete examples from our events, I will share how story-driven discussions can make technical and professional topics more approachable, and how communities can create inclusive spaces where people from different backgrounds and experience levels can learn from each other.",{"title":3358,"describe":3361,"type":30},"I run \"Engineer Anime（エンジニアニメ）,\" a community where engineers share lessons gained from anime. Using these insights as a starting point, we explore themes such as organizational design, architecture, and careers.\r\n\r\nGrounded in shared narratives, the community creates dialogue across domains and experience levels, encouraging cross-disciplinary exchange.\r\n\r\nIn this session, I will share how a community built around lessons from anime enables inclusive, domain-spanning interaction.",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FXRJJV3",{"id":3365,"room":3366,"start":95,"end":1155,"language":70,"track":3367,"speakers":3369,"zh":3373,"en":3376,"tags":3377,"uri":3378},"XXYQ89",{"en":522,"zh-hant":10},{"id":524,"name":3368},{"en":526,"zh-hant":10},[3370],{"id":3159,"avatar":3160,"zh":3371,"en":3372},{"name":3162,"bio":3163},{"name":3162,"bio":3163},{"title":3374,"describe":3375,"type":30},"Porting baremetal Go and Rust to the Aspeed AST2700","This talk will explain and showcase running Go code directly on the silicon of the AST2700 chip. This includes the troubles and blockers during porting and how I overcame them with a full hardware demo. I will briefly go over baremetal Tamago\u002FGo, how it works and how it can be utilized on modern platforms as well has how others can port it to different chips as well.",{"title":3374,"describe":3375,"type":30},[402],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FXXYQ89",{"id":3380,"room":3381,"start":327,"end":193,"language":328,"track":3382,"speakers":3384,"zh":3392,"en":3395,"tags":3396,"uri":3397},"YTAA3Q",{"en":1154,"zh-hant":10},{"id":330,"name":3383},{"en":332,"zh-hant":332},[3385],{"id":3386,"avatar":3387,"zh":3388,"en":3391},"YU9T3H","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FYU9T3H_mCxnFUa.webp",{"name":3389,"bio":3390},"Nirjal Bhurtel","Nirjal Bhurtel is a Computer Engineering graduate from Kathmandu University and an embedded systems engineer currently working as an Embedded Software Architect at Yatri Motorcycles. His work focuses on building communication systems, firmware, and testing infrastructure for electric vehicles, bridging the gap between hardware and software systems, and is currently leading Hardware In Loop Testing.\r\n\r\nHe previously worked as a Software Engineering Intern at Intra2Net AG, Germany, where he contributed to Linux-based system development and gained experience working with large-scale testing environments.\r\n\r\nNirjal has a strong background across embedded systems, backend development, and distributed systems, with hands-on experience in C\u002FC++, Python, Rust, and modern web technologies. He has led and contributed to multiple projects ranging from IoT systems and FPGA-based computing to data-driven applications and developer tools.\r\n\r\nHe is also an active researcher, with several publications in areas such as IoT, network security, and high-performance computing.\r\n\r\nIn addition to his engineering work, Nirjal is passionate about building practical, scalable systems and exploring better testing methodologies. He has worked with the Avocado testing framework in Linux environments and is currently exploring its adoption for hardware-in-the-loop (HIL) testing in embedded systems.",{"name":3389,"bio":3390},{"title":3393,"describe":3394,"type":30},"From Autotest to Avocado: Modernizing System Testing on Ubuntu","Autotest and KVM virt-test have long served as the backbone of system and virtualization testing in Linux environments. However, their tightly coupled architecture and operational complexity present growing challenges for modern development workflows, particularly in terms of modularity, scalability, and CI\u002FCD integration.\r\nAs highlighted in [Ubuntu Testing Automation](https:\u002F\u002Fwiki.ubuntu.com\u002FTesting\u002FAutomation), Ubuntu’s testing ecosystem has historically relied on Autotest and KVM autotest. Over time, this ecosystem has evolved, with KVM autotest and virt-test transitioning into the Avocado and Avocado-VT projects. Avocado represents a natural progression of the Autotest lineage, reflecting a broader shift toward more maintainable, extensible, and developer-friendly testing frameworks.\r\nThis talk provides a clear and technical comparison between Autotest and Avocado, focusing on their architectural differences, strengths, and trade-offs. We will explore key features of Avocado, including its Python-based test model, plugin architecture, parameterized execution, and enhanced reporting capabilities. The role of Avocado-VT as a continuation of virt-test for virtualization testing will also be examined.\r\nRather than advocating for a full rewrite, the session outlines a practical migration strategy for Ubuntu-based systems, demonstrating how teams can incrementally transition from Autotest to Avocado while preserving existing test investments. We will examine when to leverage Avocado-VT for compatibility and when to adopt native Avocado tests, along with approaches for integrating Avocado into modern CI\u002FCD pipelines.\r\nAttendees will gain a balanced understanding of both frameworks, along with actionable guidance on adopting Avocado to build more maintainable, scalable, and future-ready system testing workflows on Ubuntu.\r\n\r\nMore info: https:\u002F\u002Fevents.canonical.com\u002Fevent\u002F146\u002Fcontributions\u002F939\u002F",{"title":3393,"describe":3394,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FYTAA3Q",{"id":3399,"room":3400,"start":382,"end":453,"language":70,"track":3401,"speakers":3403,"zh":3411,"en":3414,"tags":3417,"uri":3418},"ZPBJGG",{"en":121,"zh-hant":10},{"id":733,"name":3402},{"en":735,"zh-hant":735},[3404],{"id":3405,"avatar":3406,"zh":3407,"en":3410},"AWPHCY","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FAWPHCY_TzOIdip.webp",{"name":3408,"bio":3409},"Kosuke Shimada (Engineer Cafe Community manager)","Engineer Cafe Community manager",{"name":3408,"bio":3409},{"title":3412,"describe":3413,"type":30},"Engineer Cafe 的開源接待系統:一個可複製的在地與全球協作典範","作為源自在地社群、以 Engineer Cafe 為基地的 OSS 開發具體實例,我們將介紹接待系統開源專案的案例。\r\n此專案包含用於接待導覽的「AI 虛擬人偶」,採用 TypeScript 與 LangGraph 等現代技術,並具備長期與短期記憶功能。我們將分享在地社群特有的開發過程與問題解決經驗。\r\n本場演講將是一個了解典範案例的機會:紮根在地的 OSS 活動,如何連結並貢獻於全球生態系。\r\n對於 COSCUP 的多元參與者而言,這不僅能為各自的在地社群活動與 OSS 開發帶來啟發,也是與福岡 OSS 社群具體交流的平台。",{"title":3415,"describe":3416,"type":30},"Engineer Cafe's OSS Reception System: A Reproducible Model for Local-Global Collaboration","As a concrete example of OSS development originating from the local community and based on Engineer Cafe, we will introduce a case study of a reception system open-sourcing project. \r\n\r\nThis project includes an \"AI avatar\" for reception navigation, utilizing modern technologies such as TypeScrip and LangGraph with long-term and short-term memory. We will share insights into the development process and problem-solving which is unique to the local community.\r\n\r\nThis talk session will be an opportunity to learn about a model case of how locally rooted OSS activities can connect to and contribute to the global ecosystem. \r\nFor the diverse participants of COSCUP, this will provide hints for their own local community activities and OSS development, as well as a platform for concrete interaction with Fukuoka OSS community.",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FZPBJGG",{"id":3420,"room":3421,"start":709,"end":68,"language":96,"track":3422,"speakers":3424,"zh":3432,"en":3435,"tags":3437,"uri":3438},"ZPKMGU",{"en":522,"zh-hant":10},{"id":524,"name":3423},{"en":526,"zh-hant":10},[3425],{"id":3426,"avatar":3427,"zh":3428,"en":3431},"UH8GLX","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FUH8GLX_lZ8om57.jpg",{"name":3429,"bio":3430},"紀煜豪\u002FSamba Chi","**現任神雲科技BIOS工程師**、~~前美超微BIOS工程師~~、~~前TomTom導航軟體架構師~~、~~前宏達電Android工程師~~、~~前圓剛韌體工程師~~、~~前台灣世曦前端工程師~~",{"name":3429,"bio":3430},{"title":3433,"describe":3434,"type":30},"Unlock BIOS from close to open","傳統上，系統底層的 BIOS 韌體多由少數供應商掌控。這種封閉的「黑箱」特性不僅限制了使用者的控制權，更往往隱藏著難以察覺的資安漏洞與效能瓶頸。\r\n\r\n本次演講將帶您深入探討韌體領域的革命性轉變——開源 BIOS。我們將剖析傳統閉源架構的痛點，並探討開源解決方案帶來的無限可能。透過開放的機制，我們得以打破硬體廠商的長期限制，實現更極致的開機速度、更高的程式碼透明度，以及可高度客製化的安全防護。\r\n\r\n無論您是系統開發者、資安專家還是開源愛好者，本場演講都將為您揭示如何奪回伺服器與個人電腦底層的絕對掌控權，並與全球社群攜手，共同建構一個更自由、透明且安全的開放硬體生態圈。",{"title":3433,"describe":3436,"type":30},"Traditionally, underlying system BIOS firmware has been controlled by a handful of vendors. This closed, \"black box\" nature not only restricts user control but also frequently conceals hard-to-detect security vulnerabilities and performance bottlenecks.\r\n\r\nThis talk will take you on a deep dive into a revolutionary shift in the firmware domain—Open-Source BIOS. We will analyze the pain points of traditional closed-source architectures and explore the infinite possibilities brought by open-source solutions. Through open mechanisms, we can break free from the long-standing restrictions imposed by hardware manufacturers, achieving ultimate boot speeds, enhanced code transparency, and highly customizable security defenses.\r\n\r\nWhether you are a system developer, a cybersecurity expert, or an open-source enthusiast, this presentation will reveal how to regain absolute control over the foundational layers of your servers and personal computers. Join us as we collaborate with the global community to build a freer, more transparent, and secure open-hardware ecosystem.",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FZPKMGU",{"id":3440,"room":3441,"start":382,"end":453,"language":328,"track":3442,"speakers":3444,"zh":3452,"en":3455,"tags":3458,"uri":3459},"ZVVRHF",{"en":167,"zh-hant":10},{"id":169,"name":3443},{"en":171,"zh-hant":172},[3445],{"id":3446,"avatar":3447,"zh":3448,"en":3451},"RYNSZY","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002F%E4%B8%8B%E8%BC%89_zKOUUfZ.png",{"name":3449,"bio":3450},"John Lu","John is a Senior Machine Learning Engineer from a Top-tier company, currently focused on developing NLP applications.\r\n\r\nHe is deeply motivated by challenges and tends to be excited by breaking conventional ways of thinking and doing. With prior experiences in Software Engineering, he works on combining the latest AI technology and engineering to transform challenges into practical solutions.",{"name":3449,"bio":3450},{"title":3453,"describe":3454,"type":59},"使用 PyTorch 從零量化 YOLOX","模型量化 (Model Quantization) 能在推論階段中有效降低計算量與記憶體使用量。 透過將模型權重與輸入資料轉換為低精度資料型態，不僅能加快運算速度，還能讓模型部署在嵌入式裝置。 本次演講將示範如何以 Python 實作模型量化。我們將從零開始，使用 PyTorch 對 YOLOX 物件偵測模型進行量化，帶領大家從原理到實作，將量化技術應用於開源模型，並轉化為可落地的解決方案。",{"title":3456,"describe":3457,"type":59},"Let's Quantize YOLOX from scratch with PyTorch","Model Quantization can reduce the computational and memory footprint during inference.  The weights and activations are represented with low-precision data types and can operations can be performed faster. Model Quantization allows us to run models on embedded devices. The goal of this talk is to demonstrate how Model Quantization could be implemented in a Pythonic way. To do so, we're going to quantize the YOLOX object detection model completely from scratch all using PyTorch.",[402],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FZVVRHF",[3461,3484,3501,3522,3545,3564,3587,3619,3640,3664,3688,3705,3728,3751,3773,3801,3820,3841,3862,3884,3906,3927,3946,3965,3988,4010,4030,4051,4072,4091,4117,4137,4157,4174,4193,4214,4236,4255,4278,4301,4317,4336,4357,4378,4397,4425,4444,4467,4488,4514,4533,4558,4577,4596,4615,4636,4657,4676,4697,4713,4732,4767,4786,4807,4827,4848,4863,4886,4905,4923,4938,4960,4982,5001,5020,5041,5059,5080,5101,5120,5135,5154,5169,5190,5209,5238,5263,5284,5304,5323,5342,5361,5380,5401,5421,5441,5462,5478,5499,5520,5541,5556,5577,5594,5612,5631,5646,5665,5680,5699,5724,5743,5764,5785,5804,5825,5846,5865,5884,5899,5916,5937,5955,5974,5995,6014,6033,6052,6073,6099,6119,6140,6159,6178,6199,6218,6258,6277,6298,6317,6338,6353,6372,6393,6408,6423,6438,6453,6474,6494,6513,6528,6547,6564,6583,6602,6623,6638,6657,6677,6692,6711,6731],{"id":3462,"room":3463,"start":3464,"end":3465,"language":70,"track":3466,"speakers":3470,"zh":3478,"en":3481,"tags":3482,"uri":3483},"3GQQBH",{"en":93,"zh-hant":10},"2026-08-09T14:30:00+08:00","2026-08-09T15:00:00+08:00",{"id":3467,"name":3468},533,{"en":3469,"zh-hant":3469},"Software Defined Vehicle accelerated by Open-Source",[3471],{"id":3472,"avatar":3473,"zh":3474,"en":3477},"VFNFLM","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FVFNFLM_ymo84U6.jpg",{"name":3475,"bio":3476},"TOKITA Hiroshi","TOKITA Hiroshi has been working as an embedded systems developer for 20 years.\r\nHe mainly has knowledge of in-vehicle Linux, especially in the infotainment area.\r\nHe is also involved in the Japanese translation of KiCad, and is involved in development and writing activities spanning open source, software, and hardware.\r\nIn recent years, he has been active as a member of the Zephyr project and as a maintainer of Raspberry Pi Pico and GD32.",{"name":3475,"bio":3476},{"title":3479,"describe":3480,"type":30},"Using Zephyr to Build Smaller Driver Domains for Virtualized Vehicle Systems","In Software-Defined Vehicle (SDV) architectures, virtualization is an important technique for isolating software functions. SoDeV, an SDV platform being developed by Automotive Grade Linux, uses Xen for virtualization. In Xen, a Driver Domain (DomD) owns physical devices and provides para-virtualized interfaces to other domains. Zephyr is a good fit for this role because it combines broad device support with a small footprint, enabling compact and well-isolated driver domains.\r\n\r\nIn this session, we will present our work on running Zephyr as a Xen driver domain, including current upstream progress. We will explain why smaller Zephyr-based driver domains are attractive, and discuss practical challenges that appear in real systems, such as partitioning hardware cleanly across domains, handling shared resources like clock control, and dealing with the associated challenges on the Xen side.\r\nWe will also demonstrate Zephyr’s DomD support live—expected to be upstream by the time of the session—on the Renesas\u002FRetronik Sparrow-Hawk V4H Board and Raspberry Pi 5.\r\n\r\nThrough this demonstration and discussion, attendees will see what is already working today, what benefits this approach can bring, and how the ideas may be applied in their own projects. We also hope to share the remaining open challenges with the community and encourage discussion on how Zephyr-based driver domains can grow into a practical open-source building block for future SDV systems.",{"title":3479,"describe":3480,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002F3GQQBH",{"id":3485,"room":3486,"start":3487,"end":3488,"language":96,"track":3489,"speakers":3494,"zh":3495,"en":3498,"tags":3499,"uri":3500},"3KSF3R",{"en":350,"zh-hant":10},"2026-08-09T09:45:00+08:00","2026-08-09T09:55:00+08:00",{"id":3490,"name":3491},522,{"en":3492,"zh-hant":3493},"Sustainable Agriculture Summit: Net Zero and Digital Trandition","農業永續雙軸轉型-地方實務分享會(農業開放資料社群)",[],{"title":3496,"describe":3497,"type":30},"社群介紹","議程介紹",{"title":3496,"describe":3497,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002F3KSF3R",{"id":3502,"room":3503,"start":3504,"end":3505,"language":328,"track":3506,"speakers":3508,"zh":3516,"en":3519,"tags":3520,"uri":3521},"3LHQ8S",{"en":407,"zh-hant":10},"2026-08-09T09:30:00+08:00","2026-08-09T10:00:00+08:00",{"id":410,"name":3507},{"en":412,"zh-hant":412},[3509],{"id":3510,"avatar":3511,"zh":3512,"en":3515},"HNDKWA","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FHNDKWA_D6soJLh.JPEG",{"name":3513,"bio":3514},"Sahil Khokhar","Sahil is a Frontend Tech Lead at Mercari Inc., where he leads teams building product-facing features for Mercari’s Marketplace. With prior contributions to open source projects at Wikimedia, he brings a strong foundation in collaborative development. An avid speaker at international tech conferences, Sahil frequently shares insights on frontend technologies and accessibility, advocating for inclusive and user-friendly digital experiences.",{"name":3513,"bio":3514},{"title":3517,"describe":3518,"type":30},"Build-Time Magic: How the React Compiler Supercharges Your App","React developers have long juggled React.memo, useCallback, and useMemo to squeeze performance out of their apps. Imagine a compiler doing that work for you automatically at build time. The React Compiler promises exactly this: fine grained memoization with zero code changes.\r\n\r\nIn this talk, I will demystify the React Compiler, explain how it works, and show real before-and-after results that prove the gains are tangible. You will learn how to adopt it safely, measure its impact in your own app, and avoid common pitfalls. With live coding and side-by-side demos, I will show how a single config flag can take an app from sluggish to snappy without touching a line of component code.\r\n\r\n\r\nWhy This Talk Matters:\r\n\r\n- The React Compiler has recently turned stable, teams are starting to adopt it in their codebases now.\r\n- Attendees get a clear benchmarking workflow to measure impact in their apps.\r\n- Live profiler demos reveal the compiler’s optimizations in action.\r\n\r\nKey Takeaways:\r\n\r\n- What the React Compiler does and how to enable it.\r\n- How to benchmark before\u002Fafter performance with tools like Lighthouse and React Profiler.\r\n- Real-world demos showing fewer re-renders, smaller bundles, and faster UIs.\r\n- When manual optimizations are still needed, and how to spot them.",{"title":3517,"describe":3518,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002F3LHQ8S",{"id":3523,"room":3524,"start":3504,"end":3505,"language":96,"track":3525,"speakers":3529,"zh":3537,"en":3540,"tags":3543,"uri":3544},"3PGBSN",{"en":167,"zh-hant":10},{"id":3526,"name":3527},536,{"en":3528,"zh-hant":3528},"Twinkle AI",[3530],{"id":3531,"avatar":3532,"zh":3533,"en":3536},"UCF3SD","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FUCF3SD_bTSLsaQ.webp",{"name":3534,"bio":3535},"Thomas Liang","Thomas 是一名跨足工程與設計領域的 ML Engineer，熱衷於探索機器學習底層技術。他目前是繁體中文開源語言模型社群 Twinkle AI 的活躍貢獻者，同時也是社群吉祥物「石虎」的創作者。作為「熬夜書坊（Late-Night Study Session）」的主持人，他致力於將艱澀的 AI 理論轉化為高重現性且易於上手的互動式教材。平時亦參與 Hugging Face Transformers、marimo 等開源專案，期望透過設計思維與「做中學（Hands-on）」，推動台灣在地生成式 AI 生態的發展。",{"name":3534,"bio":3535},{"title":3538,"describe":3539,"type":30},"從共讀到開源共創：Twinkle AI 熬夜書坊的 LLM 實戰與在地化模型應用","大語言模型（LLM）技術日新月異，單靠閱讀論文或文件已經難以跟上發展速度，真正的理解往往建立在「親手把玩程式碼」之上。\r\n\r\n本議程將分享 Twinkle AI 開源社群中「熬夜書坊（Late-Night Study Session）」的營運經驗與技術結晶。我們將開源專案 TwinkleAI-LLM-Book-Club 作為核心載體，帶領會眾一窺如何將重量級著作《Hands-On Large Language Models》轉化為社群共學的開源教材。\r\n\r\n演講中將展示我們如何利用開源的互動式 Notebook 環境（如 Jupyter 甚至具備高重現性的環境），帶領參與者一步步拆解 Tokenization 的編解碼過程、解析 Embeddings 向量魔法，並實際串接 Twinkle AI 專屬的繁中在地化模型（如 gemma-3-4B-T1-it）進行推論。這不僅是一場關於 LLM 底層技術的分享，更是展示如何透過開源協作與做中學（Hands-on），降低技術門檻，凝聚台灣在地 AI 生態圈的實踐指南。",{"title":3541,"describe":3542,"type":30},"From Co-reading to Co-creation: Hands-on LLMs and Localized Models at Twinkle AI Late-Night Study Session","As Large Language Models (LLMs) evolve rapidly, true understanding comes from hands-on coding rather than just reading papers. This session shares the technical insights and community experiences from Twinkle AI's \"Late-Night Study Session.\" Using our open-source repository, TwinkleAI-LLM-Book-Club, based on \"Hands-On Large Language Models,\" we will demonstrate how to transform a heavyweight book into interactive, community-driven materials.\r\n\r\nWe will walk through the mechanics of Tokenization and Embeddings, and showcase inferences using Twinkle AI's localized Traditional Chinese model (gemma-3-4B-T1-it). This session is a practical guide on lowering the barrier to entry through open-source collaboration and empowering the local AI ecosystem in Taiwan.",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002F3PGBSN",{"id":3546,"room":3547,"start":3548,"end":3549,"language":96,"track":3550,"speakers":3552,"zh":3558,"en":3561,"tags":3562,"uri":3563},"3VVWJU",{"en":350,"zh-hant":10},"2026-08-09T14:00:00+08:00","2026-08-09T14:40:00+08:00",{"id":3490,"name":3551},{"en":3492,"zh-hant":3493},[3553],{"id":3554,"avatar":3,"zh":3555,"en":3557},"HAZPP7",{"name":3556,"bio":10},"橘子會吃醋",{"name":3556,"bio":10},{"title":3559,"describe":3560,"type":30},"我在農業現場：永續傳承的橘子園","一個關於橘子會吃醋的小故事：善待環境＋善待果樹＝美好果實。土地不會說話，來自橘子會吃醋的柑橘，沒有亮麗外表，但顆顆飽滿多汁，這就是土地回饋友善的人們的心意。由於橘子會吃醋的張老闆從鬼門關前走一遭，了解健康的重要。因此放下空調業，返鄉務農，堅持善待的栽種柑橘，以醋浸泡蚵殼粉及天然礦物，取代化學催熟劑及生長激素，找回果實原有風味。關心消費者的健康， 從「心」開始。",{"title":3559,"describe":3560,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002F3VVWJU",{"id":3565,"room":3566,"start":3567,"end":3568,"language":96,"track":3569,"speakers":3571,"zh":3579,"en":3582,"tags":3585,"uri":3586},"7AP8HY",{"en":263,"zh-hant":10},"2026-08-09T13:00:00+08:00","2026-08-09T13:30:00+08:00",{"id":44,"name":3570},{"en":46,"zh-hant":47},[3572],{"id":3573,"avatar":3574,"zh":3575,"en":3578},"WCJNBL","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FWCJNBL_vWL9xs9.webp",{"name":3576,"bio":3577},"NZ","蘇恩立（NZ），現就讀於國立臺灣師範大學資訊工程學系三年級。對資訊安全與網路治理深感興趣，目前負責維運臺灣學術網路（TANet）首個 Tor 節點。\r\n\r\nEn-Li Su (NZ) is a junior student in the Department of Computer Science and Information Engineering at National Taiwan Normal University (NTNU). He has a deep interest in cybersecurity and Internet governance, and is currently responsible for maintaining the first Tor node on the Taiwan Academic Network (TANet).\r\n\r\nGitHub : https:\u002F\u002Fgithub.com\u002Fsu-nz\r\nInstagram : https:\u002F\u002Fwww.instagram.com\u002Fsu_n_z",{"name":3576,"bio":3577},{"title":3580,"describe":3581,"type":30},"在學校種洋蔥？臺師大 Tor 節點建立實務與 EFF Tor University Challenge 經驗談","學術網路因其頻寬穩定且 IP 聲譽良好，一直是全球 Tor 網路中最重要的支柱之一。然而，在校園環境中部署 Tor 節點常面臨技術配置與校內政策的雙重挑戰。\r\n\r\n本演講將分享講者在國立臺灣師範大學（NTNU）建立學術 Tor 節點的完整歷程，包含：\r\n\r\n- 計畫緣起：介紹 EFF 發起的 Tor University Challenge 及其對全球匿名網路的意義。\r\n- 技術實踐：基於 Linux 環境的節點配置、防火牆規則設定...。\r\n- 行政協調：如何與學校資訊單位溝通，以及如何應用 EFF 提供的法律指南。\r\n- 社群影響：建立節點後的觀察、流量分析與對後續校園推廣的建議。\r\n\r\n希望能藉此經驗，鼓勵更多臺灣學研單位參與全球隱私基礎設施的建設。",{"title":3583,"describe":3584,"type":30},"Growing Onions at School: Building Tor Relays at NTNU and the EFF Tor University Challenge","University networks are the backbone of a resilient Tor ecosystem. Yet, bringing Tor into the classroom is easier said than done. Between technical hurdles and administrative red tape, how can students successfully deploy a node?\r\n\r\nIn this talk, I will walk you through the full lifecycle of the NTNU Tor Project, including:\r\n\r\nThe Mission: Why we joined the EFF Tor University Challenge to support global digital rights.\r\n\r\nThe Tech Stack: A deep dive into Linux-based node setup and securing the network within a campus backbone.\r\n\r\nThe \"Human\" Firewall: How to navigate university politics, talk to IT centers, and handle abuse reports using EFF's legal templates.\r\n\r\nThe Data: Real-world traffic analysis from an academic relay and what it means for the future of campus privacy.\r\n\r\nOur goal is to provide a reproducible roadmap for other universities in Taiwan to join the global fight for privacy and internet freedom.",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002F7AP8HY",{"id":3588,"room":3589,"start":3590,"end":3591,"language":96,"track":3592,"speakers":3596,"zh":3611,"en":3614,"tags":3617,"uri":3618},"7EEV38",{"en":522,"zh-hant":10},"2026-08-09T12:10:00+08:00","2026-08-09T12:40:00+08:00",{"id":3593,"name":3594},517,{"en":3595,"zh-hant":10},"Open Source Inspired Hardware",[3597,3604],{"id":3598,"avatar":3599,"zh":3600,"en":3603},"7MRPLD","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002F7MRPLD_uSF7yy4.webp",{"name":3601,"bio":3602},"莊易騰","畢業於國立成功大學電機工程研究所，主要研究領域為 RISC-V CPU Design、處理器微架構設計與 performance modeling。研究期間專注於 RISC-V 處理器核心、全系統模擬平台、RTL 驗證流程與快取\u002F記憶體系統效能評估，並具備從指令集模擬器、RTL Core、Verilator\u002FSystemC 模擬環境，到 Linux\u002FFreeRTOS 開機與 co-simulation debug 流程的實作經驗。\r\n\r\n對開源硬體、RISC-V 生態系與處理器設計工具鏈有高度興趣，持續關注 XiangShan、BOOM、CVA6 等開源處理器專案，並希望透過實作與分享，讓更多人了解如何從軟體、硬體與系統層級切入 CPU 設計與效能分析。",{"name":3601,"bio":3602},{"id":3605,"avatar":3606,"zh":3607,"en":3610},"KGHQ8G","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FKGHQ8G_vFcHWmo.webp",{"name":3608,"bio":3609},"蕭珮珊","我是國立成功大學電機工程研究所碩士生，研究方向聚焦於電腦架構、RISC-V 設計以及 CPU 微架構最佳化研究。\r\n我目前於電腦架構與系統實驗室（Computer Architecture and System Laboratory, CASLab）進行研究。\r\n\r\n目前我的主要研究內容包括 RISC-V CPU 微架構設計、RTL 實作，以及 Linux／FreeRTOS 移植，並著重於前端與記憶體子系統最佳化，例如分支預測與快取微架構。\r\n我透過架構層級的設計探索、效能分析，以及硬體／軟體協同設計技術，致力於提升系統效率。",{"name":3608,"bio":3609},{"title":3612,"describe":3613,"type":30},"從零開始建立 RISC-V 全系統模擬平台","受到香山處理器、BOOM、CVA6 等開源處理器專案啟發，我們從零開始建立一套 RISC-V 全系統模擬平台，目標是支援處理器設計、軟體部署、RTL 驗證與效能探索。本平台包含可開啟 Linux 與 FreeRTOS 的 RISC-V 指令集模擬器，並已成功整合兩顆自研 RTL RISC-V 核心：一顆可開啟 Linux 的 6 級順序處理器，以及一顆可開啟 FreeRTOS 的 11 級順序處理器。 \r\n\r\n在系統層級上，我們使用 Verilator 將 RTL Core 編譯為 C++ 模型，並透過 SystemC 建立 Bus 與週邊設備，使 RTL 能在 C++ 層級進行全系統模擬，並可在一小時內完成開機。為了加速處理器開發，我們也建立 Co-simulation 機制，在 Core 提交指令時即時與指令集模擬器比對架構狀態，使錯誤能在發生當下停止模擬並輸出除錯資訊。 \r\n\r\n本演講將分享這個平台的架構設計、Linux\u002FFreeRTOS 開機經驗、RTL co-simulation debug 流程，以及如何用 C++ 實作可調整大小、路數、hit latency、replacement policy 與 prefetcher 的高效能非阻塞快取模型。我們也會展示如何透過 Konata 觀察核心與快取管線化狀態，並整合 DRAMSys 進行更接近真實系統的效能評估。",{"title":3615,"describe":3616,"type":30},"Building a Full-System RISC-V Simulation Platform from Scratch","Inspired by open-source processor projects such as XiangShan, BOOM, and CVA6, we built a RISC-V full-system simulation platform from scratch. The goal of this platform is to support processor design, software deployment, RTL verification, and performance exploration. It includes a RISC-V instruction set simulator capable of booting Linux and FreeRTOS, and has successfully integrated two in-house RTL RISC-V cores: a 6-stage in-order processor capable of booting Linux, and an 11-stage in-order processor capable of booting FreeRTOS.\r\n\r\nAt the system level, we use Verilator to compile the RTL core into a C++ model, and use SystemC to build the bus and peripheral devices. This allows the RTL design to perform full-system simulation at the C++ level and successfully boot within one hour. To accelerate processor development, we also built a co-simulation mechanism. Whenever the core commits an instruction, the architectural state is compared against the instruction set simulator in real time. When an error occurs in the core, the simulation can stop immediately and provide debugging information.\r\n\r\nIn this talk, we will share the architecture of the platform, our experience booting Linux and FreeRTOS, the RTL co-simulation debugging flow, and how we implemented a high-performance non-blocking cache model in C++ with configurable cache size, associativity, hit latency, replacement policy, and prefetcher. We will also demonstrate how Konata can be used to observe the pipelined behavior of the core and cache, and how DRAMSys is integrated to perform performance evaluation under conditions closer to a real system.",[402],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002F7EEV38",{"id":3620,"room":3621,"start":3622,"end":3623,"language":328,"track":3624,"speakers":3626,"zh":3634,"en":3637,"tags":3638,"uri":3639},"7JPPXJ",{"en":1253,"zh-hant":10},"2026-08-09T13:35:00+08:00","2026-08-09T14:05:00+08:00",{"id":901,"name":3625},{"en":903,"zh-hant":904},[3627],{"id":3628,"avatar":3629,"zh":3630,"en":3633},"KRN3BV","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FKRN3BV_PFgoi8k.jpg",{"name":3631,"bio":3632},"Kan (Koala)","在資訊管理與計算機科學之間持續漂流探索的無尾熊。從最初投入資訊系統開發與資料分析實作，逐步轉向深度學習與電腦視覺研究，並透過參與開源社群與實務專案，將研究成果延伸至醫療影像與工業視覺檢測等真實應用場景。期望藉由開源協作與人工智慧技術的結合，推動模型從論文走向工程現場，真正解決現實世界中的複雜問題。\r\n\r\nA koala continuously drifting and exploring between the fields of Information Management and Computer Science. Starting from hands-on work in information system development and data analytics, the journey gradually shifted toward research in deep learning and computer vision. Through active involvement in open-source communities and practical engineering projects, research outcomes have been extended to real-world application domains such as medical image analysis and industrial visual inspection. By integrating open collaboration with artificial intelligence technologies, the aim is to bridge the gap between academic models and engineering deployment, enabling AI systems to address complex challenges in real-world environments.",{"name":3631,"bio":3632},{"title":3635,"describe":3636,"type":30},"From Annotation Tools to Real-World AI: My Open-Source Journey with Labelme, Medical Imaging, and Industrial Vision","In practical artificial intelligence development, while model design is important, project success often depends more on data annotation workflows, toolchain integration, and engineering deployment. This talk draws on my real-world contributions to the open-source annotation tool Labelme, showing how practical user needs led to multilingual localization and feature improvements, as well as the technical trade-offs involved in collaborating with upstream maintainers and the community. Building on this experience, the session explores how open-source tools help bridge AI research and real-world applications, highlighting challenges such as data quality and privacy in medical image segmentation and data distribution shifts and real-time requirements in multi-site industrial vision deployment. Through cross-domain case studies, it emphasizes the critical role of open-source tooling in modern AI workflows and demonstrates how addressing real problems through community collaboration can effectively support engineering practice in medical and industrial AI systems.",{"title":3635,"describe":3636,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002F7JPPXJ",{"id":3641,"room":3642,"start":3643,"end":3549,"language":96,"track":3644,"speakers":3649,"zh":3656,"en":3659,"tags":3662,"uri":3663},"7TEGJH",{"en":9,"zh-hant":10},"2026-08-09T14:10:00+08:00",{"id":3645,"name":3646},535,{"en":3647,"zh-hant":3648},"Let's Read the Source Code","帶您讀源碼",[3650],{"id":3651,"avatar":3652,"zh":3653,"en":3655},"EPWFM8","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FEPWFM8_pp4DUJZ.jpg",{"name":3296,"bio":3654},"自稱雜學系工程師。\r\n名片上寫 iOS 工程師，技能樹卻點滿了 React 和 Vue。\r\n有時候會忍不住手癢，跑去寫點 Go 和 Rust。\r\n希望用自己的產品，為這個世界添一點新鮮感。",{"name":3296,"bio":3654},{"title":3657,"describe":3658,"type":30},"帶上你的 Coding Agent，我們現場替 Zed Editor 送一個 PR 吧","打開大型開源專案，面對一整個陌生的 repository、龐雜的程式碼與交錯複雜的 dependency，熱情往往瞬間冷掉。\r\n\r\n這場演講中，我會帶著 Coding Agent 走進 Zed Editor 的 codebase，從零開始理解問題、追查 bug、梳理脈絡、提出修改，並在台上一路推進到送出 Pull Request。整個過程不只呈現成功片段，而是盡可能還原真實開發現場：如何探索陌生專案、如何與 AI 一起釐清線索、如何判斷它的推論是否可靠，以及當它失誤時，如何把方向重新拉回來。\r\n\r\n如果你曾想貢獻開源卻不知道從哪裡開始，或你也好奇 AI Coding 工具在真實、混亂、不可預期的工程情境中究竟能幫上多少忙，這場演講會給你一個沒有濾鏡的答案，以及一套可以立刻帶回去實踐的方法。",{"title":3660,"describe":3661,"type":30},"Bring Your Coding Agent: Let’s Live Submit a PR to Zed Editor","Opening a large open source project can be intimidating: an unfamiliar repository, thousands of lines of code, and a web of dependencies are often enough to drain away the initial excitement.\r\n\r\nIn this session, I will bring a Coding Agent into the Zed Editor codebase and show, from scratch, how to understand the problem, investigate a bug, trace the surrounding logic, propose a fix, and work all the way toward submitting a Pull Request live on stage. Rather than showing only polished success cases, I will present the process as it actually happens: how to explore an unfamiliar project, how to collaborate with AI to narrow down clues, how to judge whether its reasoning is trustworthy, and how to steer it back when it goes off track.\r\n\r\nIf you have ever wanted to contribute to open source but did not know where to start, or if you are curious about how AI coding tools perform in real, messy, and unpredictable engineering environments, this talk will offer an unfiltered look at the process and a practical workflow you can take home and apply immediately.",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002F7TEGJH",{"id":3665,"room":3666,"start":3667,"end":3668,"language":70,"track":3669,"speakers":3674,"zh":3682,"en":3685,"tags":3686,"uri":3687},"87EJBH",{"en":282,"zh-hant":10},"2026-08-09T10:45:00+08:00","2026-08-09T11:15:00+08:00",{"id":3670,"name":3671},531,{"en":3672,"zh-hant":3673},"Google Developer Party","GDG TW (Google Developers Groups Taiwan)",[3675],{"id":3676,"avatar":3677,"zh":3678,"en":3681},"38ZMFE","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002F38ZMFE_sXGvYc6.png",{"name":3679,"bio":3680},"Sasha Denisov","Sasha is CTO at Brainform.ai with over 20 years of experience architecting scalable enterprise systems. With a strong engineering background, his expertise spans frontend, backend, cloud infrastructure, mobile development, and AI — from cloud-based generative AI to on-device solutions. He specializes in building robust, production-ready products using a variety of technologies and frameworks. Sasha has delivered solutions across fintech, digital media, and entertainment. He is a Google Developer Expert for Cloud, AI, Firebase, Flutter, and Dart, co-organizes the Flutter Berlin Community, and is a recognized international speaker and writer, having presented at 30+ conferences worldwide.",{"name":3679,"bio":3680},{"title":3683,"describe":3684,"type":30},"Evolution of flutter_gemma: From a Simple Wrapper to a Cross-Platform AI Engine","flutter_gemma started as a thin wrapper over MediaPipe for running Google Gemma on Android and iOS. Today it's a full-fledged platform for on-device AI: 6 platforms (Android, iOS, Web, macOS, Windows, Linux), 2 inference engines, multimodal support, function calling for local agents, and on-device RAG with vector search. The plugin has been featured by Google AI for Developers.\r\nIn this talk, we'll walk through the evolution via key engineering decisions. How adding Web platform forced us to rethink file handling and led to sealed classes instead of string URLs. Why we had to spin up a gRPC server in Kotlin with a bundled JVM for desktop — and how to automate the build through Xcode build phases. How the Strategy pattern allowed us to add a second inference engine (LiteRT-LM) without rewriting existing code, and how the Adapter pattern helped reuse the MediaPipe implementation. Why we needed chunk buffering for engines with fundamentally different APIs.\r\nWe'll dive deep into on-device RAG: how to build a SQLite VectorStore that works identically on mobile and in the browser via WASM. I'll show real bugs and their fixes — from Web hot restart crashes to iOS Simulator limitations with vision models.\r\nThe latest addition is genkit_flutter_gemma — a bridge to Google's Genkit for Dart that enables hybrid AI pipelines. Now you can seamlessly combine on-device inference with cloud-based models in a single Genkit flow: run lightweight tasks locally for speed and privacy, and escalate complex reasoning to the cloud — all orchestrated through one unified pipeline.\r\nEvery architectural decision in flutter_gemma is an answer to a specific problem. Minimum theory, maximum code, diagrams, and stories of \"how it broke and why it looks like this now\".",{"title":3683,"describe":3684,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002F87EJBH",{"id":3689,"room":3690,"start":3691,"end":3692,"language":328,"track":3693,"speakers":3695,"zh":3699,"en":3702,"tags":3703,"uri":3704},"8AZJYT",{"en":1253,"zh-hant":10},"2026-08-09T09:00:00+08:00","2026-08-09T09:10:00+08:00",{"id":1258,"name":3694},{"en":1260,"zh-hant":1261},[3696],{"id":1264,"avatar":1265,"zh":3697,"en":3698},{"name":1267,"bio":10},{"name":1267,"bio":10},{"title":3700,"describe":3701,"type":30},"COSCUP x UbuCon Asia 2026: Welcome Day 2","開場",{"title":3700,"describe":3701,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002F8AZJYT",{"id":3706,"room":3707,"start":3708,"end":3709,"language":42,"track":3710,"speakers":3712,"zh":3720,"en":3723,"tags":3726,"uri":3727},"8KYD9D",{"en":237,"zh-hant":10},"2026-08-09T10:10:00+08:00","2026-08-09T10:40:00+08:00",{"id":241,"name":3711},{"en":243,"zh-hant":243},[3713],{"id":3714,"avatar":3715,"zh":3716,"en":3719},"BHDNXN","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002F2024_AVATAR_RYUDOAWARU_el2c827.jpg",{"name":3717,"bio":3718},"Mu-Fan Teng","滕牧帆 (Mu-Fan Teng \u002F Owaru Ryudo)，Ruby Taiwan 總召，寫 Ruby 第 19 年。白天寫 Rails，晚上和週末把 Ruby 推到 web app 以外的地方 — 攤位遊戲、教學工具、IoT 玩具、ruby.wasm 實驗。常在 COSCUP、SITCON、RubyConf TW 的 Ruby 攤位後面負責解釋「為什麼你應該寫 Ruby」。相信讓人寫到第一行 Ruby，比讓人讀第一篇 Ruby 文章重要。\r\n\r\nMu-Fan Teng (Owaru Ryudo) is the Chief Organizer of Ruby Taiwan and a 19-year Rubyist. Rails by day; by night and weekend, he pushes Ruby into places it doesn't belong yet — booth games, teaching tools, IoT toys, ruby.wasm experiments. You'll find him behind the Ruby booth at COSCUP, SITCON, and RubyConf TW, explaining why you should write Ruby. He believes getting someone to write their first line of Ruby matters more than getting them to read their first Ruby article.",{"name":3717,"bio":3718},{"title":3721,"describe":3722,"type":30},"你的 ruby.wasm 在手機上其實會壞","SITCON 2026 我們在 Ruby Taiwan 攤位放了一個現場用 Ruby 寫一行程式的小遊戲：Jekyll 靜態站 + ruby.wasm，在瀏覽器裡跑 Ruby 4.0。設計上很單純，工程上完全不是。\r\n\r\n上線第一天我們發現：iOS Safari 整個卡住、Android Chrome 靜默 hang 在載入畫面、學生切到別的 app 回來題目就消失。最後這個跑了一千人的攤位，背後其實是三個讓我們改寫一半程式碼的手機瀏覽器地雷。\r\n\r\n這場 talk 會帶你進到這個專案的引擎室：我們怎麼把 EXPECTED_MAP 整段砍掉、改用 `rubyVM.eval` + StringIO + `srand(seed)` 讓答案在前端即時算；中、英、日三語的 i18n 怎麼從 Liquid 一路注入到 JavaScript；7 題標準 Ruby 加 3 題 JS／Python／PHP 對照題的跨語言流程怎麼設計；以及那三個手機 bug 真正的 root cause 與修法。\r\n\r\n適合對 ruby.wasm、前端 Ruby、或「怎麼把 Ruby 推到 Rails 以外」有興趣的所有人。\r\n\r\n---\r\n\r\nAt SITCON 2026, the Ruby Taiwan booth ran a live coding game: a Jekyll static site + ruby.wasm 4.0, with students writing one-line Ruby in the browser. Simple in concept, brutal in practice.\r\n\r\nDay one, we found out: iOS Safari froze, Android Chrome silently hung at load, students who switched apps lost their progress. The booth that served 1000 students was held together by fixes for three mobile-browser landmines.\r\n\r\nThis talk takes you into the engine room: how we replaced a giant EXPECTED_MAP with `rubyVM.eval` + StringIO + `srand(seed)` to compute answers in the browser; how zh\u002Fen\u002Fja i18n flows from Liquid into JavaScript; how the 7 Ruby + 3 cross-language (JS\u002FPython\u002FPHP) format works; and the real root cause behind each mobile bug.\r\n\r\nFor anyone curious about ruby.wasm, Ruby in the browser, or pushing Ruby beyond Rails.",{"title":3724,"describe":3725,"type":30},"Your ruby.wasm Doesn't Quite Work on Mobile","At SITCON 2026, the Ruby Taiwan booth ran a live coding game: a Jekyll static site + ruby.wasm 4.0, with students writing one-line Ruby in the browser. Simple in concept, brutal in practice.\r\n\r\nDay one, we found out: iOS Safari froze, Android Chrome silently hung at load, students who switched apps lost their progress. The booth that served 1000 students was held together by fixes for three mobile-browser landmines.\r\n\r\nThis talk takes you into the engine room: how we replaced a giant EXPECTED_MAP with `rubyVM.eval` + StringIO + `srand(seed)` to compute answers in the browser; how zh\u002Fen\u002Fja i18n flows from Liquid into JavaScript; how the 7 Ruby + 3 cross-language (JS\u002FPython\u002FPHP) format works; and the real root cause behind each mobile bug.\r\n\r\nFor anyone curious about ruby.wasm, Ruby in the browser, or pushing Ruby beyond Rails.",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002F8KYD9D",{"id":3729,"room":3730,"start":3505,"end":3732,"language":1256,"track":3733,"speakers":3737,"zh":3745,"en":3748,"tags":3749,"uri":3750},"8LVKCA",{"en":3731,"zh-hant":10},"TR214","2026-08-09T10:30:00+08:00",{"id":3734,"name":3735},523,{"en":3736,"zh-hant":3736},"Golang TW x Cloud Native",[3738],{"id":3739,"avatar":3740,"zh":3741,"en":3744},"D8JJXV","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FD8JJXV_Dqks4uQ.webp",{"name":3742,"bio":3743},"hsin","信誠金融科技(Footprint-AI)創辦人與執行長\r\n\r\n畢業於成功大學資訊工程研究所，擁有深厚的系統架構實力。曾作為新加坡數位保險獨角獸公司(IglooInsure)的首位員工與技術主管，從零開始建構支撐高併發業務的技術架構。身為開源專案的長期貢獻者，他擅長利用 Go 語言與系統容器技術解決複雜的基礎設施問題。現為 Footprint AI 創辦人，發起開源專案 Containarium，致力於透過技術創新實現雲端開發環境的成本民主化與自動化。\r\n\r\nFounder & CEO of Footprint-AI\r\n\r\nGraduated with an M.S. in Computer Science from National Cheng Kung University (NCKU), HsinHo Yeh is a seasoned technical leader and the Founder of Footprint AI. He was the first employee and Head of Engineering at a prominent Digital Insurance Unicorn in Singapore, where he spearheaded the architecture for hyper-growth systems. As an active open-source contributor, he specializes in Go and system container technologies. Through his latest project, Containarium, he aims to revolutionize cloud development infrastructure by making it more cost-effective and resilient for teams worldwide.",{"name":3742,"bio":3743},{"title":3746,"describe":3747,"type":30},"From 1 to 100: Building a Cost-Efficient Cloud Dev Environment with LXC and Containarium","在雲端開發盛行的時代，為每位開發者配置獨立 VM 既昂貴又浪費資源。Containarium 是一個基於 Go 語言與 LXC 技術的開源專案，旨在解決這個痛點。它能在單一 VM 上快速佈署數百個具備完整 OS 體驗、支援 SSH 存取、且能執行 Docker 的隔離開發環境。\r\n\r\n本場次將分享我們如何利用 LXC 系統容器取代傳統 VM，並透過 SSH Jump Host 與雲端原生自動化技術，在不犧牲安全性的前提下，將基礎設施成本降低 90% 以上。我們也將介紹 Containarium 如何在 Spot Instance（競價實例）上實現自動恢復，打造穩定的開發環境。",{"title":3746,"describe":3747,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002F8LVKCA",{"id":3752,"room":3753,"start":3754,"end":3755,"language":96,"track":3756,"speakers":3758,"zh":3766,"en":3769,"tags":3771,"uri":3772},"8VHXKE",{"en":192,"zh-hant":10},"2026-08-09T11:30:00+08:00","2026-08-09T12:00:00+08:00",{"id":196,"name":3757},{"en":198,"zh-hant":198},[3759],{"id":3760,"avatar":3761,"zh":3762,"en":3765},"VBCW8T","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FVBCW8T_byuIUvx.webp",{"name":3763,"bio":3764},"吳啟聖 Tim Wu","Tim Wu is a senior NPU architect and compiler engineer at a Taiwan-based AI chip startup (Kneron). He specializes in AI compiler infrastructure and software frameworks for next-generation NPU chips targeting large model inference at embedded system costs.",{"name":3763,"bio":3764},{"title":3767,"describe":3768,"type":30},"c4c: the smallest full feature C++ compiler for x86, arm and RISC-V based xPU","當前主流編譯器，真的適合 xPU 嗎？\r\n\r\n以 Google TPU、Meta MTIA、Tesla Dojo 等為代表的新一代 AI 加速器，多採用 RISC-V + Tensor Core 的混合架構，試圖在靈活性與性能之間取得平衡。表面上，這讓開發者可以沿用 LLVM 等成熟工具鏈，避免從零打造編譯器。\r\n\r\n但實務上，問題剛好相反。\r\n\r\nxPU 的效能關鍵，往往來自對硬體細節的極致掌控：memory layout、DMA、pipeline、register usage——這些都難以透過通用編譯器自動最佳化。開發者最終仍回到大量 inline assembly，與工具鏈「對抗」，而非協作。\r\n\r\n這場分享將從這個落差出發，重新思考編譯器的角色：如果最佳化邏輯本質上是 hardware-specific，是否應該被重新定義？\r\n\r\n在被 LLVM 折磨兩年之後，我做了一個選擇——既然大型語言模型已經能生成編譯器，那為什麼不從零開始，打造一個真正理解 xPU 的 C++ 編譯器？",{"title":3767,"describe":3770,"type":30},"Are modern compilers really designed for xPU?\r\n\r\nNew-generation AI accelerators like Google TPU, Meta MTIA, and Tesla Dojo increasingly adopt a hybrid architecture—RISC-V combined with Tensor Cores—to balance flexibility and performance. In theory, this allows developers to reuse mature toolchains like LLVM\u002FMLIR, avoiding the need to build custom compilers from scratch.\r\n\r\nIn practice, the opposite happens.\r\n\r\nThe performance-critical path on xPU depends heavily on hardware-specific optimizations: memory layout, DMA orchestration, software pipelines, and register control. These are difficult for general-purpose compilers to infer. Developers inevitably fall back to handwritten inline assembly—fighting the toolchain instead of leveraging it.\r\n\r\nThis talk explores that gap and rethinks the role of the compiler. If optimization is inherently hardware-specific, should the compiler be redesigned around that reality?\r\n\r\nAfter spending two years wrestling with LLVM, I made a decision: if modern LLMs can already write C compilers, why not build one that actually understands xPU?\r\n\r\nThis is not just about building a new toolchain—it’s about redefining what a compiler should be.",[402],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002F8VHXKE",{"id":3774,"room":3775,"start":3776,"end":3777,"language":96,"track":3778,"speakers":3780,"zh":3793,"en":3796,"tags":3799,"uri":3800},"97QGUT",{"en":326,"zh-hant":10},"2026-08-09T15:10:00+08:00","2026-08-09T15:25:00+08:00",{"id":3268,"name":3779},{"en":3270,"zh-hant":3271},[3781,3788],{"id":3782,"avatar":3783,"zh":3784,"en":3787},"JEVXQA","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FJEVXQA_Gnt6P4q.webp",{"name":3785,"bio":3786},"廖庭暘","來自蘭陽平原的小小工程師，目前任職於永豐金，對前後端開發都有些經驗，喜歡腦動大開做一些看起來很沒用，實際上也沒什麼用，但很有趣的事情",{"name":3785,"bio":3786},{"id":3789,"avatar":3,"zh":3790,"en":3792},"UE9HJ3",{"name":3791,"bio":10},"張琪鈺",{"name":3791,"bio":10},{"title":3794,"describe":3795,"type":30},"異議喵！你的冷知識來源不足：用 AI Agent 打造冷知識審判與卡牌對戰系統","曾幾何時，你有沒有想過「畫大餅」的由來可追溯到三國時期？魏國盧毓曾以「名如畫地作餅，不可啖也」比喻空有名聲、不能實用。原來一句職場吐槽，也藏著意想不到的冷知識。\r\n\r\n在資訊爆炸的時代，碎片化知識往往轉身即忘。為了讓學習與記憶變得更有趣，我們開發了開源趣味專案——逆轉喵判：一款結合 AI Agent、冷知識審判、卡牌對戰與桌面虛擬萌寵的智慧互動系統。\r\n\r\n使用者在日常與 Agent 互動時，系統會根據對話補上一則相關冷知識，並延伸小問題；答對後可獲得貓飼料，用來餵養桌面貓，提升牠的飽食度與好感度。隨著好感度變化，貓貓會解鎖不同外觀、動作與互動反應，讓碎片知識透過養成回饋被重新記住。\r\n\r\n為避免冷知識淪為隨機產生的有趣廢話，逆轉喵判以工程化方法建立評分流程，透過來源查證、語意比對與另一個 LLM 裁判，從冷度、可信度、關聯度、驚奇度、趣味度與可傳播度建立六邊形評分，將冷知識轉化為可收藏、可出戰的「知識喵牌」。\r\n\r\n玩家可帶著日常獲得的知識喵牌，透過另一套對戰系統，與未知挑戰者展開冷知識決鬥；桌面貓也會成為戰鬥夥伴，牠的飽食度、好感度與心情將影響對戰中的分數權重、技能觸發率與情境加成。本場分享將拆解 Prompt Engineering、Skill 設計、幻覺降低、LLM 裁判與遊戲化對戰機制。",{"title":3797,"describe":3798,"type":30},"Meowjection! — An AI-powered trivia court where cats judge, cards battle, and facts fight back.","Have you ever wondered where the phrase “drawing a big pie” comes from? Its origin can be traced back to the Three Kingdoms period, when Lu Yu of the Wei state used the phrase “a name is like drawing a cake on the ground — it cannot be eaten” to describe something that sounds impressive but has no practical value. In other words, even a casual workplace complaint may hide an unexpected piece of trivia.\r\n\r\nIn an age of information overload, fragmented knowledge is often forgotten almost instantly. To make learning and memory more engaging, we created an open-source playful project: Meowjection! — an intelligent interactive system that combines AI Agents, trivia judgment, card battles, and a desktop virtual pet.\r\n\r\nDuring everyday interactions with the Agent, the system provides not only the original response, but also a contextually relevant trivia fact and a follow-up question. When users answer correctly, they earn cat food to feed their desktop cat, improving its fullness and affection levels. As affection increases, the cat unlocks different appearances, animations, and interactive reactions, turning fragmented knowledge into a memorable pet-raising experience.\r\n\r\nTo prevent trivia from becoming random “interesting nonsense,” Meowjection! introduces an engineering-driven evaluation pipeline. Through source verification, semantic comparison, and a separate LLM judge, each trivia fact is scored across six dimensions: Novelty, Factuality, Relevance, Surprise, Interestingness, and Shareability. The result is a hexagonal rating system that transforms trivia into collectible and playable Knowledge Meow Cards.\r\n\r\nPlayers can bring the Knowledge Meow Cards they earn from daily interactions into trivia battles against unknown challengers. The desktop cat also becomes a battle companion: its fullness, affection, and mood influence scoring weights, skill trigger rates, and contextual bonuses during battles. In this talk, we will break down the implementation behind Prompt Engineering, Skill design, hallucination reduction, LLM judging, and gamified battle mechanics.",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002F97QGUT",{"id":3802,"room":3803,"start":3464,"end":3465,"language":42,"track":3804,"speakers":3806,"zh":3814,"en":3817,"tags":3818,"uri":3819},"9QK8SH",{"en":1154,"zh-hant":10},{"id":286,"name":3805},{"en":288,"zh-hant":288},[3807],{"id":3808,"avatar":3809,"zh":3810,"en":3813},"ZWKNG8","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FZWKNG8_ogLBH78.webp",{"name":3811,"bio":3812},"Cary Huang","Born in Taiwan, Cary Huang is a recognized PostgreSQL Significant Contributor and the author of the SynchDB PostgreSQL extension. With over 7 years of experience working across multiple areas of the PostgreSQL ecosystem, his work includes distributed databases, serverless database architecture, PostgreSQL internals enhancements, data security, replication technologies, and open source tooling.\r\n\r\nCary is also actively involved in the global PostgreSQL open source community through patch contributions, reviews, blogs, meetup events, conference volunteering, PostgreSQL training, and technical talks. \r\n\r\nBefore Highgo, he worked in smart metering energy sector, designing innovative and scalable solutions for energy utilities for 7 years. He holds a Bachelor’s degree in Electrical Engineering from the University of British Columbia (UBC) since 2012, and has extensive expertise in networking, database internals, security, smart metering innovations, PostgreSQL databases, architecture design, as well as embedded systems.\r\n\r\nWith a strong passion for open-source technologies and high-performance database systems, Cary continues to contribute to PostgreSQL development and create innovative designs.",{"name":3811,"bio":3812},{"title":3815,"describe":3816,"type":30},"My Journey from First Patch to PostgreSQL Significant Contributor","Over the past 7 years, I contributed to the Global PostgreSQL Open-Source community through patches, patch reviews, technical blogs, conferences, and community participation. What started as small contributions gradually became deeper involvement in PostgreSQL core development and the open-source community.\r\n\r\nIn 2025, my patch featuring Parallel TID Range Scan was accepted into the upcoming PostgreSQL 19 that makes me recognized as a PostgreSQL Significant Contributor in the global community. This talk shares the lessons, challenges, and experiences from that journey, along with tips and insights for Taiwanese PostgreSQL developers or users interested in contributing to Global PostgreSQL Open Source community.",{"title":3815,"describe":3816,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002F9QK8SH",{"id":3821,"room":3822,"start":3823,"end":3824,"language":328,"track":3825,"speakers":3827,"zh":3835,"en":3838,"tags":3839,"uri":3840},"9SSDAR",{"en":899,"zh-hant":10},"2026-08-09T11:10:00+08:00","2026-08-09T11:40:00+08:00",{"id":330,"name":3826},{"en":332,"zh-hant":332},[3828],{"id":3829,"avatar":3830,"zh":3831,"en":3834},"LDKAH3","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FLDKAH3_91hCdt0.webp",{"name":3832,"bio":3833},"Ananthu C V","Debian Developer, Linux Kernel Engineer at Qualcomm,Cinephile, Weeb, Bookworm; and interested in virtually everything.",{"name":3832,"bio":3833},{"title":3836,"describe":3837,"type":30},"IT@School, or how we renounced the proprietary and chose free software","In the southern state of Kerala in India, every government school runs a distribution named KITE (previously IT@School), based on ubuntu, in IT labs. Thanks to this, we grew up being familiar with linux and ubuntu, although we were not aware of the importance of all this as students. This will be an introduction to IT@School, and the importance of free software education from a young age.\r\n\r\nMore info: https:\u002F\u002Fevents.canonical.com\u002Fevent\u002F146\u002Fcontributions\u002F955\u002F",{"title":3836,"describe":3837,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002F9SSDAR",{"id":3842,"room":3843,"start":3709,"end":3823,"language":96,"track":3844,"speakers":3846,"zh":3854,"en":3857,"tags":3860,"uri":3861},"9U9TCT",{"en":167,"zh-hant":10},{"id":3526,"name":3845},{"en":3528,"zh-hant":3528},[3847],{"id":3848,"avatar":3849,"zh":3850,"en":3853},"BFNYPY","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FBFNYPY_g3FRlCk.webp",{"name":3851,"bio":3852},"Haley","我是一名 AI 工程師，目前專注於 LSTM 時序預測的開發。我具備電機與腦科學背景，研究所期間專攻利用 fMRI 記錄腦部狀態，並嘗試透過 GAN 技術重建受試者當下的視覺情景。 由於自身深為拖延症與思緒過載所苦，我發起了「零樹計畫（ZeroTree）」，期望藉由遊戲化技能樹的設計來輔助自我成長。專案初期，我將《實現：達成目標的心智科學》\r\n 一書中的心理學概念實作為目標設立系統；近期更挑戰在兩週的極短時間內從零開始，將 Twinkle AI 台灣原生模型落地整合進專案，打造出以「心情日記」為核心的在地化 AI 陪伴系統。我熱衷於將生硬的技術結合心理學，轉化為有溫度的日常實用工具。",{"name":3851,"bio":3852},{"title":3855,"describe":3856,"type":30},"讓台灣原生 LLM 長出在地語氣：用 Twinkle AI 4B 打造「星光型小零」","ZeroTree 是一個結合心理學與遊戲化技能樹的目標管理專案，最初是我為了解決自身拖延症與思緒過載而打造。本議程將分享我如何以極限的「兩週挑戰」為起點，從零開始將 Twinkle AI 專為台灣語境打造的 4B 模型（gemma-3-4B-T1-it）落地整合到 ZeroTree 中，以「心情日記」為核心，打造出具有開源社群與鄉民幽默感的「星光型小零」AI 陪伴原型，【並歷經數月持續打磨迭代】。這是一場本地小模型（SLM）【從原型到落地】的實戰紀錄。我將深入分享在本地端整合時遇到的四大挑戰與解法：包含如何解決小模型容易照抄 JSON 範例的格式問題、如何優化 System Prompt 避免動態資訊導致 CPU 推理延遲（提高 KV Cache 命中率）、如何在無獨顯的筆電上利用量化與 UX 設計彌補等待感，以及不綁定帳號的樹洞型專案該如何設計匿名隱私防護機制。",{"title":3858,"describe":3859,"type":30},"Growing Local Voices in Native LLMs: Building the \"Starlight Zerobot\" with Twinkle AI 4B","ZeroTree is a gamified goal-management project combining psychology and skill trees, initially created to overcome my own procrastination. This session follows my journey from a 2-week extreme integration challenge — bringing Twinkle AI's Taiwan-native 4B model (gemma-3-4B-T1-it) into ZeroTree from scratch — through months of continuous iteration. Centered around a \"mood diary,\" I developed the \"Starlight Zerobot,\" an AI companion prototype featuring Taiwanese open-source community humor and local context. I will dive deep into four major challenges and solutions of deploying a local Small Language Model (SLM): JSON formatting constraints, System Prompt optimization for CPU inference caching, UX design to mitigate hardware latency, and anonymous privacy protection for local projects.",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002F9U9TCT",{"id":3863,"room":3864,"start":3777,"end":3865,"language":96,"track":3866,"speakers":3868,"zh":3876,"en":3879,"tags":3882,"uri":3883},"A7LL8G",{"en":326,"zh-hant":10},"2026-08-09T15:40:00+08:00",{"id":3268,"name":3867},{"en":3270,"zh-hant":3271},[3869],{"id":3870,"avatar":3871,"zh":3872,"en":3875},"TAY83U","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FTAY83U_76mLza9.webp",{"name":3873,"bio":3874},"Mohuhu開發團隊","我們是一群來自金融、法律、零售、科技工程的開發者，因共同進修 AI 數據學習而集結。\r\n這份跨領域的交會，催生了我們打造「企業資訊模糊搜尋系統」的共同理念。我們致力於將複雜的 RAG 技術與多策略檢索架構整合，解決台灣在地化搜尋中常見的名稱錯置與語意斷層。對我們而言，開源不只是技術分享，更是一種實踐「資訊易得性」的方式。我們希望能消弭專業領域的資訊門檻，讓高資安要求的企業與一般使用者，都能透過這套系統，在安全的前提下實現精準、直覺的情報檢索。",{"name":3873,"bio":3874},{"title":3877,"describe":3878,"type":30},"RAG 落地實踐：用開源 LLM 打造高資安、零門檻的台灣公司情報搜尋系統","RAG（檢索增強生成）在真實生產環境的落地，常受限於企業資安紅線與硬體資源。本分享以 Mohuhu 平台為例，探討如何利用開源 LLM 與系統整合技術，解決台灣上市公司資料中複雜的語意關聯與搜尋容錯問題。\r\n\r\n我們將解析如何將 AI 能力拆解並注入系統環節：\r\n\r\n- LLM 驅動索引強化：利用 Google Gemini 作為知識 Agent，在入庫階段自動產出非結構化別名（如「護國神山」、「2330」），從根本解決向量搜尋在特定領域的語意缺失。\r\n\r\n- 四策略融合檢索（Hybrid RAG）：本系統核心為「加權評分引擎」，將檢索拆解為四個維度並合計信心分數：\r\n1. 精準比對：確保代號與統編之絕對準確。\r\n2. 中文模糊搜尋：處理名稱錯置與冗餘詞。\r\n3. 注音\u002F拼音容錯：解決輸入法未切換（如 vm6vm42j\u002F）之痛點。\r\n4. RAG 向量語意：整合本地 Ollama (bge-m3) 與 pgvector 處理深層語意。\r\n\r\n- 高資安環境部署取捨：針對離線地端場景，提出「AI 預處理、地端輕量化檢索」架構。透過 PyInstaller 封裝，讓企業在完全離線下仍能享受 LLM 強化後的體驗。\r\n\r\n- 自動化工作流：利用 OpenClaw 與 Telegram Bot 監控從爬蟲到 Embedding 的 ETL 管線，並實作失敗自動回補機制。\r\n\r\n聽眾獲益\r\n1. 混合搜尋實作：四路並行加權架構與權重調校細節。\r\n2. RAG 實戰策略：利用 LLM 預處理提升特定領域檢索召回率（Recall）。\r\n3. 工程設計模式：在資安、硬體限制與 AI 效能間取得平衡的落地經驗。",{"title":3880,"describe":3881,"type":30},"Secure RAG: Bridging LLMs and Local Deployment for Corporate Intelligence Search","Below is the English abstract for your submission, refined to reflect your technical depth and strategic engineering choices:\r\n\r\nAbstract\r\nThe real-world deployment of RAG (Retrieval-Augmented Generation) often faces challenges from corporate security policies and hardware constraints. This session uses the Mohuhu platform as a case study to explore how open-source LLMs and system integration techniques solve complex semantic relationships and search fault-tolerance issues for Taiwan’s listed company data.\r\n\r\nWe will deconstruct how AI capabilities are injected into different system layers:\r\n\r\nLLM-Driven Index Enrichment: Using Google Gemini as a knowledge agent to automatically generate unstructured aliases (e.g., \"護國神山,\" \"2330\") during the indexing stage, fundamentally addressing semantic gaps in vector search for domain-specific data.\r\n\r\nFour-Strategy Hybrid RAG: The system’s core is a weighted scoring engine that aggregates confidence scores across four dimensions:\r\n\r\nExact Match: Ensuring absolute accuracy for stock codes and tax IDs.\r\n\r\nChinese Fuzzy Search: Handling character misplacements and redundant terms.\r\n\r\nPhonetic\u002FBopomofo Fault-Tolerance: Solving search pain points caused by incorrect input methods (e.g., typing vm6vm42j\u002F).\r\n\r\nSemantic Vector Search: Integrating local Ollama (bge-m3) and pgvector to process deep semantic connections.\r\n\r\nDeployment Trade-offs for High-Security Environments: Proposing an \"AI Pre-processing + Lightweight Local Retrieval\" architecture for offline scenarios. Using PyInstaller for encapsulation, allowing enterprises to enjoy LLM-enhanced experiences in completely offline environments.\r\n\r\nAutomated LLM Workflows: Utilizing OpenClaw and Telegram Bots to monitor the ETL pipeline—from crawling to embedding—and implementing automated retry mechanisms for failures.\r\n\r\nKey Takeaways\r\nHybrid Search Implementation: Detailed design of a four-way parallel weighted architecture and weight-tuning strategies.\r\n\r\nRAG Practical Strategies: Improving retrieval recall in specific domains using LLM-based data pre-processing.\r\n\r\nEngineering Design Patterns: Practical experience in balancing data security, hardware limitations, and AI performance during deployment.",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FA7LL8G",{"id":3885,"room":3886,"start":3865,"end":3887,"language":96,"track":3888,"speakers":3890,"zh":3898,"en":3901,"tags":3904,"uri":3905},"ALLCAN",{"en":39,"zh-hant":10},"2026-08-09T16:10:00+08:00",{"id":384,"name":3889},{"en":386,"zh-hant":10},[3891],{"id":3892,"avatar":3893,"zh":3894,"en":3897},"RBVVQY","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FRBVVQY_iRvIjWK.webp",{"name":3895,"bio":3896},"陳敬霖 Sky Chen","陳敬霖（Sky），目前於國泰金控擔任後端與區塊鏈工程師。日常開發主要與 Node.js、TypeScript 及 Go 語言為伍，專注於企業級 Hyperledger Fabric 網路的建置、維運與疑難排解。除了與程式碼奮戰，我平時也是團隊中的康樂股長，喜歡主揪活動，享受與團隊及社群共同除錯、一起破關的樂趣。\r\n\r\nI, Sky Chen, am a Backend & Blockchain Engineer at Cathay Financial Holdings. Specializing in Node.js, TypeScript, and Go, I build and troubleshoot enterprise Hyperledger Fabric networks. Beyond coding, I love organizing team events and tackling challenges together with the team and community.",{"name":3895,"bio":3896},{"title":3899,"describe":3900,"type":30},"Hyperledger Fabric 深度維運與排障指南：那些官方文件沒寫的血淚經驗","Hyperledger Fabric (HLF) 擁有嚴謹的架構與豐富的官方文件，但在企業級環境實際落地與維運時，工程師往往會遇到文件上沒有寫的「驚喜」。本議程將濃縮過去一年在 HLF 專案中經歷的真實踩坑血淚史，帶領聽眾潛入 HLF 的底層架構，探索那些在 Tutorial 中學不到的實戰排障技巧。\r\n\r\n本場分享將圍繞三個真實的維運災情與架構迷思展開：\r\n- 末端憑證輪替大翻車與底層拆解：當更換末端憑證時，面臨「兩個 Channel 各自只成功更新一半」的驚險時刻。我們將從更新腳本一路往下追查，剝開底層 Envelope 結構的真實邏輯，找出背後的真相。\r\n- 從「重啟治百病」到 GitHub 發 Issue：遇到連官方文件都找不到的神祕報錯 failed determine org 怎麼辦？分享從使用重啟作為 Workaround，到最終深入追蹤原始碼、並向 HLF 官方提交 Bug 單的開源參與歷程。\r\n- 通往 Orderer 的單向道：破解 gRPC 網路連線的常見迷思，探討為何 Peer 到 Orderer 其實只需要單向連通，背後可能有什麼機制與優缺點。",{"title":3902,"describe":3903,"type":30},"What the Hyperledger Fabric Tutorials Don't Tell You","Hyperledger Fabric (HLF) boasts a robust architecture and comprehensive official documentation. Yet, when deploying and maintaining it in real-world enterprise environments, engineers often stumble upon \"surprises\" entirely missing from the manuals. This session condenses a year's worth of hard-earned lessons from real HLF project pitfalls. We will dive deep into HLF's underlying architecture to explore practical troubleshooting techniques that no tutorial will ever teach you.\r\n\r\nThis session will center around three real-world operational disasters and architectural myths:\r\n\r\nThe Certificate Rotation Catastrophe & Low-Level Deconstruction: That heart-stopping moment during an end-entity certificate rotation when \"two channels only successfully updated halfway.\" We will trace the issue from the deployment scripts all the way down to the core, unpacking the true logic of the underlying Envelope structure to uncover what really happened.\r\n\r\nFrom \"Have You Tried Restarting It?\" to Opening a GitHub Issue: What do you do when confronted with the mysterious failed determine org error that doesn't exist in the official docs? I will share the open-source contribution journey—from relying on server reboots as a temporary workaround, to diving straight into the source code, and finally submitting a bug report to the official HLF repository.\r\n\r\nThe One-Way Street to the Orderer: Debunking common myths surrounding gRPC network connections. We will explore why the connection from Peer to Orderer actually only requires a one-way path, the underlying mechanisms behind this design, and its inherent pros and cons.",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FALLCAN",{"id":3907,"room":3908,"start":3909,"end":3910,"language":42,"track":3911,"speakers":3913,"zh":3921,"en":3924,"tags":3925,"uri":3926},"ANSNYB",{"en":9,"zh-hant":10},"2026-08-09T12:50:00+08:00","2026-08-09T13:20:00+08:00",{"id":3645,"name":3912},{"en":3647,"zh-hant":3648},[3914],{"id":3915,"avatar":3916,"zh":3917,"en":3920},"AYFDBQ","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FAYFDBQ_XY9UZgE.webp",{"name":3918,"bio":3919},"Michael Yuan","Dr. Michael Yuan is a maintainer of the WasmEdge project and a co-founder of Second State. He is the author of 5 books on software engineering published by Addison-Wesley, Prentice-Hall, and O'Reilly. Michael is a long- time open-source developer and contributor. He had previously spoken in many industry conferences including Open Source Summit, RustLab Conference, KubeCon, KCD, Qcon and many more.",{"name":3918,"bio":3919},{"title":3922,"describe":3923,"type":30},"Open-Source AI Infrastructure That People Can Actually Run: Lessons from Building a Local Sovereign Cloud","Open-source projects often succeed on innovation but fail on operability: installation, upgrades, permissions, observability, and support. In this talk, we’ll walk through what it takes to turn an open-source AI infrastructure project into a system real users can run safely on their own hardware.\r\n\r\nUsing the open source AI OS Olares as the case study, we’ll cover packaging, app lifecycle, GPU support, private data access, and remote access for a local sovereign cloud. We’ll also discuss what breaks when a project moves from developer excitement to actual everyday usage, and how to design the system so contributors and users can both succeed.\r\n\r\n\r\nAttendees will leave with a clearer mental model for how to package open-source infrastructure in a way that is both technically sound and community-friendly.",{"title":3922,"describe":3923,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FANSNYB",{"id":3928,"room":3929,"start":3754,"end":3755,"language":70,"track":3930,"speakers":3932,"zh":3940,"en":3943,"tags":3944,"uri":3945},"ASC9WV",{"en":1154,"zh-hant":10},{"id":286,"name":3931},{"en":288,"zh-hant":288},[3933],{"id":3934,"avatar":3935,"zh":3936,"en":3939},"XSLRQ8","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FXSLRQ8_vTyGyxx.webp",{"name":3937,"bio":3938},"Sergey Pronin","Sergey is a product leader focusing on delivering robust open-source solutions. After having consistently contributed to the growth of Percona and its product open source culture as a VP of Product, he is now leading the OpenEverest open source database platform. Digital nomad, who enjoys spreading the word about data on Kubernetes. He loves playing tennis and hockey, but his major passions are his family, three kids and a dog.",{"name":3937,"bio":3938},{"title":3941,"describe":3942,"type":30},"When Life Gives You Containers, Make an Open Source RDS: A Kubernetes Love Story","This isn’t your typical technical talk. We’ll take you on a step-by-step adventure, starting from a humble single database in a container and adding components one by one, just like we did. You’ll witness firsthand how we tackled real-world challenges, from storage and scaling to monitoring and UI design, to create OpenEverest - CNCF Sandbox project.\r\n\r\nYou’ll walk away with a deep understanding of how Kubernetes can be used to orchestrate complex and stateful applications (like databases clusters).\r\n\r\nJoin us and discover how you can break free from vendor lock-in, save costs, and build a database that’s truly yours. This is your chance to learn from our triumphs and tribulations, and be inspired to create your own open source success story.",{"title":3941,"describe":3942,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FASC9WV",{"id":3947,"room":3948,"start":3505,"end":3732,"language":1469,"track":3949,"speakers":3951,"zh":3959,"en":3962,"tags":3963,"uri":3964},"AVUHFD",{"en":1063,"zh-hant":10},{"id":1065,"name":3950},{"en":1067,"zh-hant":1068},[3952],{"id":3953,"avatar":3954,"zh":3955,"en":3958},"RH9RMP","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FRH9RMP_t1lXyvu.webp",{"name":3956,"bio":3957},"Masaki Murakami","Masaki Murakami is an experienced open knowledge advocate and long-time contributor to the Japanese Wikipedia since 2007, serving as an administrator since 2010. Demonstrating a strong commitment to community growth, he joined the Wikimedians of Japan User Group in 2023 and currently serves as Project Manager and Acting Representative. He specializes in driving local outreach and cross-sector collaborations to foster open data ecosystems. Furthermore, Masaki contributes to the regional movement as a member of the Core Organizing Team for the upcoming ESEAP Summit 2027 in Osaka. He is dedicated to advancing knowledge equity and sustainable community governance.\r\n\r\n村上正記（Masaki Murakami）是一位致力於推廣開放知識的維基媒體人。他自2007年起開始參與日語維基百科的編輯，並自2010年起擔任管理員，長期協助社群營運。2023年，他加入「日本維基媒體用戶組」（Wikimedians of Japan User Group），現擔任專案經理及代理代表，積極推動在地推廣活動與開放數據生態系統的發展。此外，他也是即將於大阪舉辦的「2027年 ESEAP 高峰會」（ESEAP Summit 2027）核心組織團隊成員。他致力於促進知識平權，並推動亞太地區社群的可持續性治理。\r\n\r\n村上正記（Masaki Murakami）は、オープンナレッジの普及に深く携わるウィキメディアンです。2007年から日本語版Wikipediaの編集を開始し、2010年からは管理者としてコミュニティの運営を支えています。2023年からはWikimedians of Japan User Groupに参画し、現在はプロジェクトマネージャーおよび代表代行として、地域のアウトリーチ活動やオープンデータ・エコシステムの発展を牽引しています。さらに、2027年に大阪で開催される国際カンファレンス「ESEAP Summit 2027」のコア・オーガナイジング・チーム（主宰メンバー）も務めており、アジア太平洋地域における知識の公平性と持続可能なコミュニティ統治の実現に向けて尽力しています。",{"name":3956,"bio":3957},{"title":3960,"describe":3961,"type":30},"How is the Wikimedia Movement in Japan Operating? Navigating Community, Collaboration, and the AI Era","Since the launch of the Japanese Wikipedia in 2001, the Wikimedia movement in Japan has evolved into a highly unique and deeply collaborative ecosystem. Initially driven by isolated online contributors, the movement has steadily transitioned into dynamic offline community-building and strategic outreach initiatives. This 30-minute session explores the historical trajectory, distinct methodologies, and future prospects of the Japanese Wikimedia community, offering valuable insights into how open knowledge can thrive in a non-Western cultural context.\r\n\r\nA cornerstone of Japan's outreach strategy is its extensive collaboration with local institutions and cultural heritage sectors, known as GLAM (Galleries, Libraries, Archives, and Museums). The session will highlight the success of \"Wikipedia Towns\"—innovative, hyper-local edit-a-thons organized across Japan. In these events, local citizens, librarians, and government officials collaborate to digitize and document regional history, tourism, and cultural assets on Wikipedia and Wikimedia Commons. By bridging grassroots participation with institutional archiving, the Japanese movement has successfully democratized knowledge production and fostered strong local open-data ecosystems.\r\n\r\nFurthermore, the presentation will address how Japanese Wikimedians are tackling global knowledge equity challenges. Through localized adaptations of international campaigns such as \"WikiGap\" and \"Art+Feminism,\" the community is actively working to bridge the gender gap and improve the representation of marginalized histories in the Japanese-language digital space. These initiatives demonstrate how global open-source movements can be effectively contextualized to address local systemic biases.\r\n\r\nCrucially, the session will confront the pressing challenges and opportunities brought about by the rapid evolution of Artificial Intelligence (AI) and Large Language Models (LLMs). As AI tools increasingly rely on Wikipedia for training data, the role of human editors is undergoing a paradigm shift. The speaker, a seasoned administrator of the Japanese Wikipedia since 2007, will analyze what is currently happening within the Japanese Wikimedia infrastructure. The discussion will delve into how the community maintains data quality, counters misinformation, and redefines the value of human-curated open knowledge in an AI-dominated information landscape.\r\n\r\nAttendees—including open-source developers, data advocates, and community managers—will gain a comprehensive understanding of the operational mechanics of one of Wikipedia's largest non-English language editions. By attending this lecture, participants will discover actionable strategies for community engagement, cross-sector partnerships, and sustainable open-knowledge governance in the age of AI.",{"title":3960,"describe":3961,"type":30},[],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FAVUHFD",{"id":3966,"room":3967,"start":3968,"end":3969,"language":96,"track":3970,"speakers":3972,"zh":3980,"en":3983,"tags":3986,"uri":3987},"B7UYHH",{"en":282,"zh-hant":10},"2026-08-09T14:20:00+08:00","2026-08-09T14:50:00+08:00",{"id":3670,"name":3971},{"en":3672,"zh-hant":3673},[3973],{"id":3974,"avatar":3975,"zh":3976,"en":3979},"LDKQVE","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FLDKQVE_Vbuv0lD.webp",{"name":3977,"bio":3978},"Evan Lin \u002F Zona Wang","Evan Lin\r\nTech Director of LINE Taiwan Developer Relations Team\r\nCloud AI and Golang Google Developer Expert\r\n\r\nZona Wang \r\n來自 LINE Taiwan DevRel 團隊。 負責規劃與推廣開發者關係相關活動，包含技術社群合作、品牌推廣與內容策展， 持續推動 LINE 與開發者社群的連結與交流。",{"name":3977,"bio":3978},{"title":3981,"describe":3982,"type":30},"AI 時代的 Developer Experience：我們如何用 Gemini 打造 LINE Bot Developer Assistant","近年 AI coding assistant 快速發展，但對許多開發者來說，真正困難的問題仍然是：\r\n\r\n不知道該從哪開始\r\n文件太多看不完\r\nwebhook\u002Fdebug 很難排查\r\nsample code 不知道如何修改\r\n\r\n在這場分享中，我們將以「LINE Bot 開發學習流程」為案例，探討 AI 如何改善 Developer Experience。\r\n\r\n我們結合：\r\n\r\nGemini\r\nMCP Server\r\nAntigravity CLI\r\nLINE 官方文件\r\n\r\n打造一個能協助開發者學習與開發 LINE Bot 的 Developer Assistant。\r\n\r\n除了介紹技術架構外，我們也會分享：\r\n\r\nAI 如何改變開發者 onboarding\r\nDeveloper Relations 在 AI 時代的新挑戰\r\n怎麼讓 AI 不只是產生 code，而是真正幫助開發者理解與學習\r\n\r\n最後也會現場 demo：\r\n從零開始透過 AI 建立一個 LINE Bot workflow。",{"title":3984,"describe":3985,"type":30},"AI-Native Developer Experience: Building a LINE Bot Developer Assistant with Gemini","As AI coding assistants become increasingly common, many developers still struggle with the same problems:\r\n\r\nToo much documentation\r\nDifficult onboarding flows\r\nDebugging webhook and API integration issues\r\nKnowing where to start when learning a new platform\r\n\r\nIn this session, we explore how AI can improve Developer Experience by building a LINE Bot Developer Assistant powered by Gemini.\r\n\r\nUsing LINE Bot development as a real-world example, we combine:\r\n\r\nGemini\r\nMCP Server\r\nAntigravity CLI\r\nLINE official documentation\r\n\r\nto create an AI-assisted workflow that helps developers learn, build, and troubleshoot LINE Bot applications more effectively.\r\n\r\nRather than focusing only on code generation, we will discuss:\r\n\r\nHow AI changes developer onboarding\r\nThe evolving role of Developer Relations in the AI era\r\nDesigning AI workflows that help developers understand, not just copy and paste\r\n\r\nWe will also demonstrate a live workflow where AI assists developers from documentation lookup to bot implementation and debugging.\r\n\r\nThis session is designed for students, developers, and anyone interested in AI-native developer tooling and the future of developer learning experiences.",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FB7UYHH",{"id":3989,"room":3990,"start":3732,"end":3991,"language":96,"track":3992,"speakers":3996,"zh":4004,"en":4007,"tags":4008,"uri":4009},"B9RVUR",{"en":67,"zh-hant":10},"2026-08-09T11:00:00+08:00",{"id":3993,"name":3994},537,{"en":3995,"zh-hant":3995},"State of the Map Taiwan 2026 \u002F Wikidata Community Summit 2026",[3997],{"id":3998,"avatar":3999,"zh":4000,"en":4003},"WGV8UG","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FWGV8UG_165VF9w.jpg",{"name":4001,"bio":4002},"王文岳 Allen Wang","Wikidata Taiwan 臺灣維基數據社群共同發起人\r\n\r\n四處拍廟、拍墓的人；近年以維基街景踏察團為主要的社群工作，常與各社群合作舉辦各式活動。過去曾任第一屆第一期立法院開放國會委員會民間委員、國家文化記憶庫：110-111年社群經營與運作模式建置計畫資訊技術協力、社團法人台灣維基媒體協會秘書長，曾共同主持有「客家@維基」資料多元應用專案以及執行李梅樹紀念館國家文化記憶庫專案「臺灣美術巨擘的最後一程—李梅樹告別式相關文物掃描」等資料開放專案。",{"name":4001,"bio":4002},{"title":4005,"describe":4006,"type":30},"以 Wikidata 建立的開放文獻編目","Wikidata 收錄了超過 1.2 億個條目，其中最大宗的資料竟然是學術文章，比例高達百分之二十而筆立第二高內容：人，僅占百分之六。\r\n\r\n在 Wikidata 中收錄的學術文章內容，臺灣的佔比相當稀少，早在李梅樹紀念館、新樂園藝術空間以及客家開放資料等合作專案中，建立相對應主題的文獻編目收錄在 Wikidata 中也是其中一個工作目標之一，這也促進了許多 Wikidata 中臺灣主題資料同時也有對應的文獻內容可以相互連結，提高不同領域資料「偶遇性」，進一步提高文獻資料的可用性與可檢索性。\r\n\r\n本次分享將以 Lightbox 攝影圖書室 Wikidata 合作案、故宮博物院期刊文章 Wikidata 合作為案例，分享文獻編目的實務操作以及文獻資料轉進 Wikidata 的實體效益與應用可能。",{"title":4005,"describe":4006,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FB9RVUR",{"id":4011,"room":4012,"start":3668,"end":4013,"language":70,"track":4014,"speakers":4016,"zh":4024,"en":4027,"tags":4028,"uri":4029},"BE7WZN",{"en":93,"zh-hant":10},"2026-08-09T11:45:00+08:00",{"id":3467,"name":4015},{"en":3469,"zh-hant":3469},[4017],{"id":4018,"avatar":4019,"zh":4020,"en":4023},"GF7833","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FGF7833_sDfAx0H.webp",{"name":4021,"bio":4022},"Flo Wohlrab","Flo is the Head of the OpenHW Group and international not-for-Profit organization dedicated to curate and maintain high quality, industrial grade, open source RISC-V Cores. Flo is fascinated by RISC-V and help to smoothen accessibility to high quality RISC-V and help promote the RISC-V ISA worldwide. He is also one of the first RISC-V International Ambassadors. Before leading the OpenHW Group Flo was Head of Sales for a commercial RISC-V provider where he took care of the markets in Japan, Europe and Middle East.",{"name":4021,"bio":4022},{"title":4025,"describe":4026,"type":30},"Eclipse SDV: Building worldwide SDV ecosystems and why open source may not be enough","In this talk Flo will introduce the  Eclipse Foundation which has started a Software Defined Vehicle Working Group. Members like Mercedes, Cariad, Bosch, Continental, and many others are already working together on a fully open source and commercially usable SDV Software layer, enabling faster innovation and reusability of base blocks. We will introduce the Eclipse SDV initiative and why this open source approach will help the worldwide car industry to enable the Software Defined Vehicle. We will highlight the achievements of this young initiative as well as critically asking why open source may not be enough to make this worldwide collaboration happen.",{"title":4025,"describe":4026,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FBE7WZN",{"id":4031,"room":4032,"start":3505,"end":3732,"language":70,"track":4033,"speakers":4037,"zh":4045,"en":4048,"tags":4049,"uri":4050},"BFCP9S",{"en":121,"zh-hant":10},{"id":4034,"name":4035},538,{"en":4036,"zh-hant":4036},"Fediverse & Social Web",[4038],{"id":4039,"avatar":4040,"zh":4041,"en":4044},"WCJCKX","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FWCJCKX_RhLyy3Z.webp",{"name":4042,"bio":4043},"Yuki Onobuchi","Infrastructure and backend engineer based in Himeji. I studied philosophy in university, exploring human-AI interactions through a culturalist lens. My primary work involves operating healthcare and academic systems, alongside freelance Python development. I've been running the Mastodon instance \"mstdn.y-zu.org\" since high school and founded FediLUG in 2023. Driven by a commitment to open source and knowledge sharing, I aim to build communities where intellect and technology intersect through decentralized networks.",{"name":4042,"bio":4043},{"title":4046,"describe":4047,"type":30},"Decentralized by Design, Connected by Culture: Japan's Fediverse and OSS Ecosystem Through FediLUG","The Fediverse has grown into a globally distributed network of independent servers, yet its expansion is not merely a technical achievement—it is a cultural one. In Japan, the adoption of decentralized social protocols like ActivityPub has been deeply intertwined with the country's long-standing open-source software (OSS) culture, Linux user groups, and maker communities. This talk explores how Japan's Fediverse has evolved, the unique challenges it faces, and how community-driven initiatives like the Fediverse Linux Users Group (FediLUG) are shaping its future.\r\n\r\nFounded in 2023, FediLUG emerged from a simple observation: instance operators in Japan were highly skilled technically but often isolated in their stewardship roles. Unlike Western ecosystems where large organizations or well-funded nonprofits often back Fediverse projects, Japan's scene relies heavily on volunteers running small-to-mid-sized instances. FediLUG was created to bridge this gap by fostering knowledge sharing, providing operational support, and building a sustainable community around decentralized social networking. Through workshops at major OSS conferences, hands-on meetups, and open documentation initiatives, we have been cultivating a space where technical expertise meets long-term community stewardship.\r\n\r\nThis session will walk you through FediLUG's activities and the insights gained from supporting Japan's Fediverse landscape. We will cover practical topics such as onboarding new operators, implementing scalable moderation policies, and maintaining healthy federation relationships in a multilingual, multi-cultural environment. Beyond operations, we will examine the governance and social dimensions of running decentralized networks. Drawing from a culturalist perspective on technology, I will discuss how valuing human effort, dialogue, and shared responsibility can lead to more resilient communities than purely protocol-driven or market-driven models.\r\n\r\nThe Japanese Fediverse also presents unique structural characteristics: a high density of niche, hobby-driven instances; strong ties to traditional Linux and BSD communities; and a growing emphasis on interoperability and data portability. These traits offer valuable lessons for global Fediverse builders, particularly around sustainability without corporate backing, balancing decentralization with usability, and fostering cross-cultural collaboration in East Asian OSS ecosystems.\r\n\r\nAttendees will leave with actionable strategies for community building, moderation at scale, and long-term instance sustainability. Whether you are an ActivityPub implementer, an instance operator, or simply interested in how decentralized networks shape social behavior, this talk will provide a grounded, community-first perspective on the future of the social web.",{"title":4046,"describe":4047,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FBFCP9S",{"id":4052,"room":4053,"start":4054,"end":4055,"language":328,"track":4056,"speakers":4058,"zh":4066,"en":4069,"tags":4070,"uri":4071},"BGYZ3B",{"en":3731,"zh-hant":10},"2026-08-09T11:20:00+08:00","2026-08-09T11:50:00+08:00",{"id":3734,"name":4057},{"en":3736,"zh-hant":3736},[4059],{"id":4060,"avatar":4061,"zh":4062,"en":4065},"VWTTC3","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FVWTTC3_tglfwUT.webp",{"name":4063,"bio":4064},"Reza jelveh","Reza Jelveh is a solutions engineer at HAMi (project-hami.io), focusing on the mechanics, constraints, and resource allocation of AI workloads on Kubernetes.\r\n\r\nHis background spans three decades—from reverse-engineering VCRs to building systems for semiconductor testing, seismic mitigation, healthcare, and data center construction. He has served as CTO for startups and a €bn-revenue government institution, navigating everything from bare-metal hardware tradeoffs to the operational friction of legacy middleware.",{"name":4063,"bio":4064},{"title":4067,"describe":4068,"type":30},"Running Multiple AI Workloads on One GPU with HAMi: Architecture and Gotchas","GPUs are expensive. Kubernetes doesn't share them well yet, DRA is still work in progress.\r\n\r\nHAMi: Heterogeneous GPU sharing for Kubernetes.\r\n\r\nWhat I'll Tell You:\r\n\r\nHow it hijacks CUDA calls without touching your app code\r\nWhy memory isolation matters\r\nSome real production use cases\r\nReal Results: Teams cut GPU costs 40-60%\r\n\r\nWho Should Show Up: K8s operators, platform engineers, anyone watching GPUs sit idle\r\n\r\ngithub.com\u002FProject-HAMi\u002FHAMi",{"title":4067,"describe":4068,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FBGYZ3B",{"id":4073,"room":4074,"start":4054,"end":4055,"language":70,"track":4075,"speakers":4077,"zh":4085,"en":4088,"tags":4089,"uri":4090},"BGZGNA",{"en":1063,"zh-hant":10},{"id":1065,"name":4076},{"en":1067,"zh-hant":1068},[4078],{"id":4079,"avatar":4080,"zh":4081,"en":4084},"LVAK3F","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FLVAK3F_cXK0aDz.webp",{"name":4082,"bio":4083},"Atsushi Shimono (html5j)","Atsushi had been working in Mozilla community of both in Japan and global projects such as devmo (current MDN) and bugzilla since 2000, and has been active in html5j which is one of the largest Web community in Japan for both organizing conferences and providing technical talks.",{"name":4082,"bio":4083},{"title":4086,"describe":4087,"type":30},"WebXR current position and beyond to the Recommendation","WebXR is a collection of APIs to interface with mixed reality consisted with Core API and feature specific modules, and enables you to provide 3D experience by managing the process of rendering 3D context, handing inputs from mixed reality hardware, and providing domain specific feature such as anchor or layers for projecting scenes.\r\nSince 2018, set of specifications are developed under the Immersive Web groups of the W3C, having a Candidate Recommendation status, and implemented in Chrome and various 3D headset devices.\r\n\r\nWebXR APIs has been evolving to enable new hardware and sensing features, and to introduce new software features for both making widely used concepts into core specification and reducing pain points to use APIs in real space.\r\n\r\nIn this talk, following brief introduction and recent updates of WebXR specifications including bringing specifications to the W3C Recommendation, I will pick three emerging topics in WebXR field in detail.\r\n\r\nFirst is 3D rendering processing, initially WebXR was built over WebGL to generate stream of 3D view, but direct use of GPU is desired and now under implementation.\r\nSecond is spatial context, which enables you to connect physical space with virtual space, such as placing and handling virtual objects in physical space as if handling as physical objects over multiple mixed reality hardware, or shaing physical positions among devies.\r\nThrid is model element to bring 3D assets as core media in the Web.",{"title":4086,"describe":4087,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FBGZGNA",{"id":4092,"room":4093,"start":3548,"end":3464,"language":96,"track":4094,"speakers":4096,"zh":4109,"en":4112,"tags":4115,"uri":4116},"BJ78VU",{"en":326,"zh-hant":10},{"id":3268,"name":4095},{"en":3270,"zh-hant":3271},[4097,4104],{"id":4098,"avatar":4099,"zh":4100,"en":4103},"YYXZZY","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FYYXZZY_fRNHwoy.jpg",{"name":4101,"bio":4102},"Samson","現任國泰金控助理架構師，擁有 9 年以上系統架構與後端開發經驗，橫跨金融、政府、醫療與車用等多個產業領域。目前職責著重於企業 AI 賦能與架構治理，推動內部 AI 導入策略與工程實踐。\r\n\r\n長期關注開源技術在垂直領域的落地應用，熟悉雲端部署與分散式系統設計。\r\n\r\n本次議程以自行實作的 AlphaCouncil 開源專案為主軸，分享在金融場景中導入 Google ADK 與開源 LLM（Gemma 4）構建多代理人系統的工程實踐。\r\n\r\n---\r\nAssistant Architect at Cathay Financial Holdings with 9+ years of system architecture and backend development experience across financial services, government, healthcare, and automotive domains. Currently focused on enterprise AI adoption and architecture governance.\r\n\r\nA practitioner interested in applying open-source technology to real-world vertical domains, with background in cloud deployment and distributed systems design.\r\n\r\nThis talk is grounded in hands-on experience building AlphaCouncil: an open-source multi-agent Taiwan stock analysis framework integrating Google ADK with locally-deployed open-source LLMs (Gemma 4).",{"name":4101,"bio":4102},{"id":4105,"avatar":3,"zh":4106,"en":4108},"SG8HJL",{"name":4107,"bio":10},"陳庭宇",{"name":4107,"bio":10},{"title":4110,"describe":4111,"type":30},"AlphaCouncil：用 Google ADK ＋ 開源模型打造多代理人辯論式台股分析系統——資料在地化 × 模型在地化","GitHub 上的 ai-hedge-fund（⭐ 52k）與 TradingAgents（⭐ 50k）讓「用 LLM 分析股票」的概念引爆開源社群——但兩者的資料源與語境都以美股為主，難以直接套用於台灣市場。\r\n\r\n本議程以 AlphaCouncil 為例，分享如何以 Google ADK（Agent Development Kit）的 SequentialAgent、ParallelAgent、LoopAgent 三種原生編排原語，融合上述兩個開源框架的設計精髓，打造一套以台灣股市為第一公民的在地化多代理人分析系統。在地化發生在兩個層次：\r\n\r\n資料層：新聞改接台灣繁體中文財經媒體（Yahoo 財經、鉅亨網、經濟日報等）；籌碼面整合 TWSE OpenAPI 的融資融券與三大法人資料（外資、投信、自營商），完全免費、無需 API 金鑰；基本面納入殖利率、月營收、本淨比等台股特有指標。\r\n\r\n模型層：透過 Ollama 在本機執行開源模型（Gemma 4、Llama 3、Qwen 等），讓台股分析資料無須離開本機，符合金融場景的資料隱私需求，同時消除商業 API 費用。Gemma 4 與 Google ADK 同屬 Google 生態系，在繁體中文指令遵循與多步驟推理上表現亮眼，是本系統首選開源底座；商業 API（Gemini、Claude、GPT）則作為可選替換方案。\r\n\r\n系統架構五個階段：四位分析師並行蒐集資料、13 位投資大師（Buffett、Graham、Munger 等）以台股視角評述、看多\u002F看空研究員 LoopAgent 辯論、交易員整合提案，最後由三方風險辯論後投資組合管理人裁決。\r\n\r\n本議程聚焦三項工程重點：Gemma 4 + ADK 的整合方式、TWSE OpenAPI 免費資料的接入實作，以及如何讓開源模型以繁體中文輸出具台股語境的人格分析報告。無論你是想用開源模型打造垂直領域 Agent，還是想把熱門英文開源專案在地化為繁體中文版本，本議程將提供一個可拆解、可復用的參考實作。",{"title":4113,"describe":4114,"type":30},"AlphaCouncil: Multi-Agent Debate-Driven Taiwan Stock Analysis with Google ADK + Open-Source LLMs — Localizing Both Data and Models","Open-source projects like ai-hedge-fund (⭐ 52k stars) and TradingAgents (⭐ 50k stars) have proven the community's appetite for LLM-powered financial analysis. But both projects are US-centric—their data sources, news feeds, and financial metrics are built around American markets and don't transfer cleanly to Taiwan.\r\n\r\nThis talk introduces AlphaCouncil, a Taiwan-first multi-agent stock analysis framework built natively on Google ADK, combining investor persona agents (inspired by ai-hedge-fund) with debate-driven decision-making (inspired by TradingAgents). Localization happens at two levels:\r\n\r\nData layer — replacing US-centric sources with Taiwan equivalents:\r\n\r\nNews: Traditional Chinese financial media (Yahoo Finance TW, 鉅亨網, 經濟日報, etc.)\r\nInstitutional flows: TWSE OpenAPI for 三大法人 (foreign investors, investment trusts, dealers) and margin trading (融資融券)—free, no API key required\r\nFundamental metrics: yield (殖利率), monthly revenue (月營收), P\u002FB ratio—Taiwan-specific KPIs absent from US-oriented feeds\r\nModel layer — running open-source LLMs locally via Ollama (Gemma 4, Llama 3, Qwen, etc.):\r\n\r\nAll 13 investor persona reports generated in Traditional Chinese with Taiwan market context\r\nFinancial data never leaves the local machine—addressing privacy requirements in financial use cases\r\nGemma 4 pairs naturally with Google ADK (same ecosystem), delivering strong Traditional Chinese instruction-following and multi-step reasoning out of the box\r\nNo commercial API costs; commercial APIs (Gemini, Claude, GPT) remain an optional swap-in\r\nThe pipeline runs in five stages—analyst team, masters panel, research debate, trader, and risk debate—orchestrated with SequentialAgent, ParallelAgent, and LoopAgent primitives to ensure a deterministic flow without LLM dynamic routing.\r\n\r\nKey engineering takeaways:\r\n\r\nIntegrating Gemma 4 via Ollama into a Google ADK multi-agent pipeline\r\nIntegrating TWSE's free OpenAPI into an ADK tool layer\r\nPrompt engineering for Traditional Chinese persona-constrained output with Taiwan market context",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FBJ78VU",{"id":4118,"room":4119,"start":4120,"end":4121,"language":96,"track":4122,"speakers":4124,"zh":4131,"en":4134,"tags":4135,"uri":4136},"BQXC39",{"en":407,"zh-hant":10},"2026-08-09T13:55:00+08:00","2026-08-09T14:35:00+08:00",{"id":410,"name":4123},{"en":412,"zh-hant":412},[4125],{"id":4126,"avatar":4127,"zh":4128,"en":4130},"MXJNZ9","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FMXJNZ9_NEIza3x.png",{"name":4129,"bio":10},"Lois",{"name":4129,"bio":10},{"title":4132,"describe":4133,"type":30},"你以為是 bug，其實是 feature：TypeScript 為什麼故意不是一個安全的型別系統？","你有沒有遇過 TypeScript 明明沒報錯，但程式在 runtime 壞掉的經驗？\r\n\r\n你以為是 bug，其實是 feature。\r\n\r\nTypeScript 的官方 Non-goals 第一條就寫著：「我們不追求 sound 或可被證明正確的型別系統，我們選擇在正確性與生產力之間取得平衡。」換句話說，TypeScript 從一開始就沒打算讓你的程式完全安全。\r\n\r\n這場議程會帶你看幾個 TypeScript 刻意「不安全」的真實案例，每個案例都會解釋：為什麼 TS 團隊選擇這樣設計、代價是什麼、以及你可以怎麼防禦。\r\n\r\n最後會帶到 TypeScript 7 的 erasable syntax，說明這個語言正在回頭確認自己的邊界——理解這件事，是從「會用 TypeScript」到「真正理解 TypeScript」的分水嶺。",{"title":4132,"describe":4133,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FBQXC39",{"id":4138,"room":4139,"start":4140,"end":3643,"language":70,"track":4141,"speakers":4143,"zh":4151,"en":4154,"tags":4155,"uri":4156},"BRGCEK",{"en":282,"zh-hant":10},"2026-08-09T13:40:00+08:00",{"id":3670,"name":4142},{"en":3672,"zh-hant":3673},[4144],{"id":4145,"avatar":4146,"zh":4147,"en":4150},"CBANNT","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FCBANNT_duSSHzl.webp",{"name":4148,"bio":4149},"HyunWoo Lee","Hello, I'm HyunWoo Lee from South Korea. I'm working for Toss, the most popular banking\u002Ffin-tech application in South Korea.\r\nAlso I'm organizer of Kotlin User Groups Seoul & GDG Korea Android.\r\n\r\nCareers & Experiences\r\n- 2024.07 ~ : Android\u002FReact Native Engineer, Viva Republica(Toss)\r\n- 2025.01 ~ : Organizer, GDG(Google Developer Groups) Korea Android\r\n- 2023.10 ~ : Organizer, Kotlin User Groups Seoul\r\n- 2023.08 ~ 2024.07: Lead, GDSC(Google Developer Student Clubs) Konkuk University, South Korea\r\n- 2021.04 ~ 2023.07: Android\u002FReact Native Developer, Mathpresso Inc.(QANDA)\r\n\r\n- Speaker of DroidKnights(South Korea) 2024\r\n- Speaker of DroidKaigi(Japan) 2024\u002F2025\r\n- Maintainer, Jindong(compose-jindong\u002Fjindong)\r\n- Contributor, DroidKaigi(Japan)\u002FDroidKnights(South Korea) Conference Application",{"name":4148,"bio":4149},{"title":4152,"describe":4153,"type":30},"Jindong: Introducing Declarative Haptics in Jetpack Compose","Mobile UI development has already moved toward a declarative paradigm, but haptic feedback is still often handled through imperative, platform-specific APIs. On Android, developers need to work with timing and amplitude arrays, while iOS requires a different approach through Core Haptics. As a result, expressing even a simple vibration pattern can quickly become difficult to read, maintain, and reuse across platforms.\r\nIn this session, I'll introduce Jindong(진동, Vibration), a declarative haptic feedback library for Compose Multiplatform. Jindong lets developers describe haptic patterns using a Compose-like DSL with concepts such as Haptic, Delay, and Sequence, while the library handles platform-specific execution details for Android and iOS.\r\nWe will look at the motivation behind declarative haptics, how Jindong integrates with Compose state and lifecycle, and how this approach can make haptic feedback more expressive, reusable, and aligned with modern Compose development.",{"title":4152,"describe":4153,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FBRGCEK",{"id":4158,"room":4159,"start":3991,"end":3754,"language":42,"track":4160,"speakers":4162,"zh":4166,"en":4169,"tags":4172,"uri":4173},"BTPMFT",{"en":67,"zh-hant":10},{"id":3993,"name":4161},{"en":3995,"zh-hant":3995},[4163],{"id":2710,"avatar":2711,"zh":4164,"en":4165},{"name":2713,"bio":10},{"name":2713,"bio":10},{"title":4167,"describe":4168,"type":30},"使用強大的 JOSM - 透過 Vibe 腳本來達到批次與大量編輯開放街圖資料","少數而且有經驗的開放街圖圖客會用功能強大的 JOSM 來編輯，但 JOSM 的一些套件也許沒有經常更新或是說明文件並不足夠，特別是 Scripting 插件。而藉由時下流行的大語言模型，如 ChatGPT，講者能夠找出能夠運作的 Python 在 Java 下的 Jython 腳本程式碼，並且能夠當上驗證是否能運作。先從能夠輸出除錯訊息，弄出取代原先不會輸出訊息的 print 指令的函式開始，再來是撰寫出能夠直接處理 OSM 標籤的尋找與取代功能，模仿類似的編輯工具 Level0。這一場次將會分享目前講者的經驗，使用 JOSM Scripting 所遇到的困難與挑戰，如何讓 JOSM Scripting 再次可以運作。",{"title":4170,"describe":4171,"type":30},"Vibe the way of Using the Powerful JOSM by Scripting, Mass-modified OpenStreetMap Data","JOSM is a powerful editing tool used by small but experienced mappers. But there are some plugin either not actively maintained or well documents. especially the Scripting plugin. By using AI-assisted programming tools like ChatGPT, the presenter can make JOSM Script tools using Python core Jython function again, from printing debugging messages by adding a new print function to replace the non-function print. And also a find and replace script to mimic the Level0 text edit function by an external text editor. This session will share the presenter's struggle and experience of making the JOSM script plugin work again.\r\n\r\nGithub repository: https:\u002F\u002Fgithub.com\u002FOsmHackTW\u002FJOSM-Scripting\r\n\r\nOutline:\r\n\r\n- Introduction\r\n- Python and XML is a good pair. Why not directly edit XML using Python Beautifulsoup?\r\n- Main Target: Taiwan's address normalization\r\n- The 500 limit of Level0, and iD text field edit is not programmable\r\n- Bring back print for debugging, and also fix for the JOSM API change \r\n- The Unicode challenge\r\n- Conclusion",[402],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FBTPMFT",{"id":4175,"room":4176,"start":3824,"end":3590,"language":328,"track":4177,"speakers":4179,"zh":4187,"en":4190,"tags":4191,"uri":4192},"BUNAKD",{"en":899,"zh-hant":10},{"id":330,"name":4178},{"en":332,"zh-hant":332},[4180],{"id":4181,"avatar":4182,"zh":4183,"en":4186},"QKNRVF","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FQKNRVF_gOFLxE8.webp",{"name":4184,"bio":4185},"Rudra Saraswat","I am the project lead & maintainer of blendOS, an immutable and atomic Linux distribution, alongside having served as an Ubuntu member and the project lead of Ubuntu Unity.",{"name":4184,"bio":4185},{"title":4188,"describe":4189,"type":30},"Implementing immutability & atomicity within desktop Linux","*This is an advancement on a presentation I had previously delivered with regards to atomic Linux distributions found [here](https:\u002F\u002Fevents.canonical.com\u002Fevent\u002F89\u002Fcontributions\u002F459\u002F).*\r\n\r\nOver the past couple of years, immutable & atomic Linux distributions have made significant strides in day-to-day usability and practicality; yet, there remains comparatively little in the way of low-level explanations with regards to how these mechanisms are implemented under-the-hood, in spite of the extent to which such attributes are desired on other, existing distributions.\r\n\r\nThe intention of this presentation, then, is to demonstrate one such implementation of atomicity within a desktop Linux distribution I maintain - blendOS - and explore how we went about implementing support for varying distribution bases (among them Ubuntu), alongside how one may develop such an implementation for their own distributions, in an attempt to ease existing systems into such a paradigm.",{"title":4188,"describe":4189,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FBUNAKD",{"id":4194,"room":4195,"start":3708,"end":3709,"language":96,"track":4196,"speakers":4198,"zh":4206,"en":4209,"tags":4212,"uri":4213},"CERYEC",{"en":39,"zh-hant":10},{"id":384,"name":4197},{"en":386,"zh-hant":10},[4199],{"id":4200,"avatar":4201,"zh":4202,"en":4205},"9G7V9L","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002F9G7V9L_3P4FcLd.webp",{"name":4203,"bio":4204},"CC Wang","嗨～我是 CC",{"name":4203,"bio":4204},{"title":4207,"describe":4208,"type":30},"劇情懶人包──關於後量子交易簽章（PQTS）的社群討論","自從 ML-DSA 在 2024 年 8 月定案後，經過一陣子議題發酵與發展，我們陸續見到幾條區塊鏈開始進入製作 POC 與效能評估的階段。這場演講將分享近期後量子交易簽章（PQTS）的社群討論，檢視各區塊鏈如何在簽名大小、狀態膨脹、驗證成本及遷移路徑之間取得平衡，並以簡顯易懂的方式介紹數個較知名的解決方案。",{"title":4210,"describe":4211,"type":30},"TL;DR: The Community Discussion Around Post-Quantum Transaction Signatures","When NIST finalized ML-DSA as FIPS 204 in August 2024, it gave the cryptography world its first standards-blessed answer to \"what replaces ECDSA on the day a Cryptographically Relevant Quantum Computer (CRQC) actually shows up?\" Gradually, the conversation has moved from mailing-list theory into working code, with several chains entering the proof-of-concept and performance-benchmarking phase and turning what used to be a thought experiment into a concrete engineering problem.\r\nThis talk is a TL;DR on where the community discussion around Post-Quantum Transaction Signatures (PQTS) actually stands today. We'll walk through several of the better-known proposals in plain language, looking at how each chain is navigating the trade-offs between signature size, state bloat, verification cost, and migration path. The goal isn't to crown a winner, but to leave you with a clear mental map of the design space and the open questions worth following.",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FCERYEC",{"id":4215,"room":4216,"start":3776,"end":3865,"language":328,"track":4217,"speakers":4219,"zh":4230,"en":4233,"tags":4234,"uri":4235},"CKHDBZ",{"en":899,"zh-hant":10},{"id":330,"name":4218},{"en":332,"zh-hant":332},[4220,4223],{"id":175,"avatar":176,"zh":4221,"en":4222},{"name":178,"bio":179},{"name":178,"bio":179},{"id":4224,"avatar":4225,"zh":4226,"en":4229},"CWZGGA","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FCWZGGA_dbD59b0.webp",{"name":4227,"bio":4228},"Yuning Liang","Yuning Liang is the Founder and CEO of DeepComputing, focusing on developing innovative technology products based on RISC-V SoMs. From the world's first RISC-V development laptop DC-ROMA to pads, workstations, remote-controlled cars, drones, and more, all are based on RISC-V chips.\r\nThe world's first RISC-V laptop, the world's first RISC-V pad capable of making phone calls, and so on, are all Yuning's masterpieces. Yuning's innovation and pioneering spirit in the RISC-V field have enabled him to create several world firsts, leading DeepComputing to gain widespread recognition in the global RISC-V product commercialization field, contributing significantly to the advancement and progress of RISC-V technology.\r\nYuning's career has taken him from the UK to Switzerland, then to South Korea, and finally to China. He has a strong practical background in embedded systems, platform APIs, and system software. In 2024, he was honored with the \"RISC-V Community Contributor Award\" and recognized as a \"Ubuntu Summit Contributor,\" further solidifying his influence in the technology sector.",{"name":4227,"bio":4228},{"title":4231,"describe":4232,"type":30},"Bringing AI to RISC-V: From Edge Devices to Developer-Ready Systems","As AI workloads continue to expand beyond the cloud, developers are increasingly looking for open, efficient, and flexible platforms to run models at the edge and on personal devices. RISC-V, as an open instruction set architecture, is emerging as a promising foundation for this shift.\r\nIn this session, Martin will share practical engineering experiences from building AI-capable RISC-V systems — covering key areas such as CPU–accelerator integration, software stack enablement, and optimization for real-world workloads.\r\nThe talk will explore how Linux-based environments can support AI development on RISC-V today, what challenges remain across toolchains and runtimes, and how upstream collaboration is helping accelerate ecosystem maturity.\r\n\r\nMore info: https:\u002F\u002Fevents.canonical.com\u002Fevent\u002F146\u002Fcontributions\u002F952\u002F",{"title":4231,"describe":4232,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FCKHDBZ",{"id":4237,"room":4238,"start":3505,"end":3709,"language":96,"track":4239,"speakers":4241,"zh":4249,"en":4252,"tags":4253,"uri":4254},"CLL3BU",{"en":544,"zh-hant":10},{"id":546,"name":4240},{"en":548,"zh-hant":548},[4242],{"id":4243,"avatar":4244,"zh":4245,"en":4248},"Q8QYXV","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FQ8QYXV_Ai8L1LZ.webp",{"name":4246,"bio":4247},"貝坦","嗨我貝坦\r\n是個對資安有興趣的資安研究員",{"name":4246,"bio":4247},{"title":4250,"describe":4251,"type":30},"：好的 正在幫您下載 VScode 惡意擴充套件｜從 GlassWorm 看開發環境供應鏈攻擊","2025 年 10 月，一隻名為 GlassWorm 的惡意蠕蟲病毒悄悄出現在數個 VS Code 的擴充套件市集，截至今日，該攻擊早已不斷演進並擴散至不同環境，累計入侵數百個套件與 repo ——攻擊對象不是伺服器，是各位開發者！\r\n\r\n本議程將帶各位聽眾拆解 GlassWorm 的實際樣本與公開分析攻擊手段與手法，逐一解析它每一個工程決策背後的邏輯——不僅從一個開發者也看得懂，更是從開發者的角度來解剖這隻蠕蟲！",{"title":4250,"describe":4251,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FCLL3BU",{"id":4256,"room":4257,"start":4258,"end":4259,"language":96,"track":4260,"speakers":4262,"zh":4270,"en":4273,"tags":4276,"uri":4277},"CNNASG",{"en":263,"zh-hant":10},"2026-08-09T10:55:00+08:00","2026-08-09T11:25:00+08:00",{"id":44,"name":4261},{"en":46,"zh-hant":47},[4263],{"id":4264,"avatar":4265,"zh":4266,"en":4269},"BJYRYX","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FBJYRYX_FgY4Oq8.webp",{"name":4267,"bio":4268},"Pellaeon Lin","Digital security researcher and trainer.\r\n\r\nI care about digital rights and FOSS.\r\n\r\nI'm user of: Linux, Firefox, LineageOS and many other FOSS.",{"name":4267,"bio":4268},{"title":4271,"describe":4272,"type":30},"以 OpenWRT 等開源軟體建立家用網路環境","本議程將介紹 OpenWRT ，一套開源的路由器作業系統。我將跳過基本的網路設定，介紹一般家用路由器不提供，但可以利用一點 OpenWRT 設定就做到的各種方便、安全、隱私功能。\r\n\r\n- 挑選支援 OpenWRT 的硬體\r\n- VLAN 、訪客網路及危險設備隔離\r\n- 多個 WAN （白話：如何把電信商給你的多個 IP 用好用滿）\r\n- 策略路由、site-to-site VPN\r\n- 危險 IP 及域名阻擋\r\n- 特殊上游網路：VPN 供應商、TOR",{"title":4274,"describe":4275,"type":30},"Build your home network with OpenWRT","This is an introduction to OpenWRT, a FOSS home router operating system. I will skip over basic network settings and focus on special things that OpenWRT enables you to do in a home network.\r\n\r\n- Picking the right OpenWRT hardware\r\n- VLAN, guest network, isolating untrusted devices\r\n- Multiple WANs\r\n- Policy-based routing, site-to-site VPN\r\n- Blocking risky IP and domains\r\n- Special WAN configuration: VPN and TOR",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FCNNASG",{"id":4279,"room":4280,"start":4281,"end":4282,"language":42,"track":4283,"speakers":4285,"zh":4293,"en":4296,"tags":4299,"uri":4300},"CTTGPG",{"en":1154,"zh-hant":10},"2026-08-09T13:45:00+08:00","2026-08-09T14:15:00+08:00",{"id":286,"name":4284},{"en":288,"zh-hant":288},[4286],{"id":4287,"avatar":4288,"zh":4289,"en":4292},"NB3FXJ","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FNB3FXJ_7oSDxUD.webp",{"name":4290,"bio":4291},"Grant Zhou","Grant Zhou is an open source database community organizer and PostgreSQL ecosystem contributor based in Metro Vancouver, Canada. He supports Highgo in open source community and ecosystem operations, organizes HOW (ivorysql.io), contributes to PGCONF.Asia, and serves as Secretary General of the China PostgreSQL Association.\r\n\r\nHe is involved in PostgreSQL-related open source projects such as IvorySQL and SynchDB, focusing on open source database adoption, Oracle-to-PostgreSQL migration scenarios, documentation, education, and cross-region community collaboration.\r\n\r\n中文简介：Grant Zhou 长期参与 PostgreSQL 与开源数据库社区建设，关注 Oracle 到 PostgreSQL 的迁移场景、数据库兼容性实践，以及亚洲社区与全球 PostgreSQL 生态之间的协作。",{"name":4290,"bio":4291},{"title":4294,"describe":4295,"type":30},"Building a PostgreSQL Open Source Community in China: From Local Users to Global Collaboration","This talk shares practical experiences from building a PostgreSQL open source community in China and connecting local users with the global PostgreSQL ecosystem.\r\n\r\nIt will cover community meetups, bilingual content, PostgreSQL education, university collaboration, documentation translation, and downstream open source projects such as IvorySQL and SynchDB. The talk will also discuss how real production pain points can be collected, anonymized, and turned into reproducible technical cases for community discussion and improvement.\r\n\r\nThe session is intended for PostgreSQL users, open source developers, educators, and community organizers. Audience members will learn practical ways to grow a local open source database community, support cross-language collaboration, and build bridges between regional communities and global upstream projects.",{"title":4297,"describe":4298,"type":30},"在中国建设 PostgreSQL 开源社区：从本地用户到全球协作","PostgreSQL 是全球最成功的开源数据库项目之一，但建设一个健康的本地社区并不只是技术问题，也需要翻译、教育、线下活动、会议、用户案例，以及连接本地用户与全球上游社区的桥梁。\r\n\r\n本演讲将分享在中国建设和支持 PostgreSQL 相关开源社区的实践经验，包括本地 Meetup、双语社区内容、PostgreSQL 教育、大学合作，以及 IvorySQL、SynchDB 等下游开源项目。\r\n\r\n演讲还将讨论如何收集并匿名化真实生产环境中的问题，将其转化为可复现的技术案例；如何通过社区活动连接用户、开发者和贡献者；以及区域社区如何更积极地参与全球 PostgreSQL 生态。\r\n\r\n本演讲适合开源社区组织者、PostgreSQL 用户、开发者、教育工作者，以及所有对跨语言、跨地区开源社区建设感兴趣的人。",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FCTTGPG",{"id":4302,"room":4303,"start":4282,"end":4304,"language":70,"track":4305,"speakers":4307,"zh":4311,"en":4314,"tags":4315,"uri":4316},"CV9JMX",{"en":1253,"zh-hant":10},"2026-08-09T14:45:00+08:00",{"id":901,"name":4306},{"en":903,"zh-hant":904},[4308],{"id":3676,"avatar":3677,"zh":4309,"en":4310},{"name":3679,"bio":3680},{"name":3679,"bio":3680},{"title":4312,"describe":4313,"type":30},"The Evolution of AI Agents in E-Commerce: From Chatbots to Autonomous Transactions","Most AI solutions in e-commerce today are still rule-based chatbots. The next step — intelligent assistants powered by RAG and function calling, capable of working with product catalogs and guiding buyers through purchases in a conversation — is gaining traction but remains far from mainstream adoption. And on the horizon, a third model is already taking shape: fully agentic commerce, where a buyer's AI agent independently discovers, compares, and purchases products without ever visiting a store's website.\r\nWhat makes this shift possible is open source. The entire infrastructure for agentic commerce is being built on open standards and protocols licensed under Apache 2.0. First came universal protocols — Anthropic's MCP for agent-to-service interaction and Google's A2A for agent-to-agent communication. Then, specialized commerce protocols emerged on top of them: Google's UCP backed by Shopify, Stripe, Visa, and 20+ partners, and OpenAI's ACP co-developed with Stripe. All open source. All community-governed.\r\nIn this talk, we'll explore where the industry currently stands on this evolutionary ladder, how the approaches of key players differ, why they chose open source as the foundation for commercial infrastructure, which early experiments have already failed — and what sellers and developers should be doing today to prepare for a world where the storefront is a context window.",{"title":4312,"describe":4313,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FCV9JMX",{"id":4318,"room":4319,"start":3991,"end":3754,"language":70,"track":4320,"speakers":4322,"zh":4330,"en":4333,"tags":4334,"uri":4335},"DL8SSQ",{"en":121,"zh-hant":10},{"id":4034,"name":4321},{"en":4036,"zh-hant":4036},[4323],{"id":4324,"avatar":4325,"zh":4326,"en":4329},"URKEZT","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FURKEZT_snQqsLz.webp",{"name":4327,"bio":4328},"Jihyeok Seo","Jihyeok from South Korea, building web services for the public good.\r\n\r\nMy personal website: https:\u002F\u002Fji.hyeok.org",{"name":4327,"bio":4328},{"title":4331,"describe":4332,"type":30},"From 0 to Fediverse","[Oeee Cafe](https:\u002F\u002Foeee.cafe) is an oekaki service for the 2020s.\r\n\r\n[Naru](https:\u002F\u002Fnaru.pub) is like Neocities, but for Koreans.\r\n\r\nI launched these two projects without any social networking functions, but they now ship with a Fediverse experience.\r\n\r\nI'll be outlining how I started these web services, and how it gained Fediverse capabilities.",{"title":4331,"describe":4332,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FDL8SSQ",{"id":4337,"room":4338,"start":3590,"end":3591,"language":96,"track":4339,"speakers":4341,"zh":4349,"en":4352,"tags":4355,"uri":4356},"DRAXDG",{"en":9,"zh-hant":10},{"id":3645,"name":4340},{"en":3647,"zh-hant":3648},[4342],{"id":4343,"avatar":4344,"zh":4345,"en":4348},"YRUXTJ","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FYRUXTJ_A8A7tYP.webp",{"name":4346,"bio":4347},"曾偉誠","CCU OS Lab",{"name":4346,"bio":4347},{"title":4350,"describe":4351,"type":30},"分層剖析 Linux Kernel 如何觀測記憶體存取行為並推論 Working Set","隨著 AI、雲端服務與大型應用持續提升記憶體需求，記憶體占用已不只是容量問題，也直接關係到成本與效能。不論是在個人電腦、行動裝置，或瀏覽器、Electron 應用等日常軟體中，「程式占用大量記憶體」也是使用者與開發者常見的困擾。\r\n\r\n然而，當我們在 Linux 上分析一個程式「用了很多記憶體」時，常會從 RSS、PSS 等指標開始觀察，但這些指標多半只能指出「目前佔用了多少記憶體」，卻難以回答「哪些 page 近期真的被存取」以及「哪些資料仍屬於程式的 working set」。\r\n\r\n因此，本議程將從 memory access profiling 的角度切入，回到 Linux kernel 的 idle page tracking、DAMON 與 `mm\u002Fworkingset.c` 之實作，分層剖析 kernel 如何透過底層訊號觀測與推論 workingset、記憶體冷熱分佈與 refault\u002Freuse 行為。最後，使用 damo、idle_page_tracking 等開源工具設計實驗展示如何對記憶體存取行為進行觀測，並估計其 working set size。",{"title":4353,"describe":4354,"type":30},"How the Linux Kernel Observes Memory Access and Infers the Working Set","As memory demands grow with AI, cloud services, and large-scale applications, memory consumption is no longer just a capacity issue—it directly impacts cost and performance. When analyzing memory usage on Linux, we typically start with metrics like RSS and PSS, but these only tell us how much memory is occupied, not which pages are actually being accessed or which data remains in a process's working set.\r\nThis talk takes a memory access profiling approach, diving into the Linux kernel's idle page tracking, DAMON, and mm\u002Fworkingset.c to dissect how the kernel observes and infers the working set, hot\u002Fcold memory distribution, and refault\u002Freuse behavior. We also demonstrate hands-on experiments using open-source tools such as damo and idle_page_tracking to observe memory access patterns and estimate working set size.",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FDRAXDG",{"id":4358,"room":4359,"start":3549,"end":3776,"language":328,"track":4360,"speakers":4362,"zh":4370,"en":4373,"tags":4376,"uri":4377},"E8EVPS",{"en":899,"zh-hant":10},{"id":330,"name":4361},{"en":332,"zh-hant":332},[4363],{"id":4364,"avatar":4365,"zh":4366,"en":4369},"ND7SB8","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FND7SB8_88ziDIX.webp",{"name":4367,"bio":4368},"Tony Yip","DevSecOps Engineer from Hong Kong living in Taipei",{"name":4367,"bio":4368},{"title":4371,"describe":4372,"type":30},"在臺灣的外籍 Linux 桌面使用者生存指南","本議程將探討居住在台北、且完全依賴 Linux 桌面環境的外籍居民與外國訪客所面臨的挑戰與困境。在臺灣，處理日常生活和行政事務往往需要與高度在地化的特定數位基礎設施進行互動。從基於晶片卡（智慧卡）的政府服務——例如全民健保（NHI）系統、網路報稅、自然人憑證——到傳統上僅支援 Windows 或 macOS 生態系的在地銀行網路 ATM（WebATM），Linux 使用者經常會遇到獨特的軟體相容性障礙。講者將分享其個人經驗，並詳細說明他們如何成功克服這些挑戰，在台灣作為 Linux 桌面使用者順利生存。\r\n\r\nMore info: https:\u002F\u002Fevents.canonical.com\u002Fevent\u002F146\u002Fcontributions\u002F947\u002F",{"title":4374,"describe":4375,"type":30},"The survive guideline for foreign Linux desktop user in Taiwan","This session discuss challenge and difficulty for expatriates and foreign visitors living in Taipei who rely exclusively on a Linux desktop environment. Navigating daily life and administrative tasks in Taiwan often requires interacting with highly specific local digital infrastructure. From smart-card-based government services, such as the National Health Insurance (NHI) system, tax filing, and Citizen Digital Certificate portals—to local banking WebATMs that traditionally favor Windows or macOS ecosystems, Linux users frequently face unique software compatibility barriers. The speaker will share their personal experience and detail how they successfully overcame these challenges to survive as a Linux desktop user in Taiwan.\r\n\r\nMore info: https:\u002F\u002Fevents.canonical.com\u002Fevent\u002F146\u002Fcontributions\u002F947\u002F",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FE8EVPS",{"id":4379,"room":4380,"start":3505,"end":3732,"language":328,"track":4381,"speakers":4383,"zh":4391,"en":4394,"tags":4395,"uri":4396},"E9LETD",{"en":899,"zh-hant":10},{"id":330,"name":4382},{"en":332,"zh-hant":332},[4384],{"id":4385,"avatar":4386,"zh":4387,"en":4390},"SJZZ3Y","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FSJZZ3Y_v3rB0OF.webp",{"name":4388,"bio":4389},"Venkatesh Chaturvedi","Venkatesh Chaturvedi works as an enterprise support engineer at Atlassian. He’s passionate about Free and Open Source Software and likes to tinker with various Linux based distributions. In his free time, he likes to explore the latest advancements in FOSS and likes to try out new and exciting tools for the CLI. He has been using Linux based distros full time for several years, for all tasks including but not limited to gaming. He currently leads the FOSS United Lucknow",{"name":4388,"bio":4389},{"title":4392,"describe":4393,"type":30},"Gaming on Linux with Ubuntu and Proton","I want to talk about how good gaming on Linux is, as compared to Windows. With the constant improvements in proton and other FOSS libraries and tools, this might just be the year of the Linux desktop.\r\n\r\nThere's been a >50% year on year growth in the number of people gaming on Linux, and this is only going to grow. The major boost came when Valve introduced the Steam Deck which basically revolutionized gaming on Linux. With the upcoming Steam Machine, we might just make up the majority of the market, overtaking Windows for good.\r\n\r\nI want to talk a bout the basic technical aspects of the entire gaming pipeline on Linux, and introduce people to some of the most important FOSS tools and libraries which help us play games on Linux. Few of these are proton, dxvk, vk3d, vk3d-proton, and I'll also be talking about how Nvidia went partially open source for Linux and why it helps the community. Might also briefly touch on the work going on with NVK and Nova.\r\n\r\nI'll also talk about the basics of how to setup an Ubuntu desktop for gaming. I'll go through the required software and tools needed to game comfortably on Ubuntu and how an end user can do this without getting into a web of complex configurations.\r\n\r\nMore info: https:\u002F\u002Fevents.canonical.com\u002Fevent\u002F146\u002Fcontributions\u002F934\u002F",{"title":4392,"describe":4393,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FE9LETD",{"id":4398,"room":4399,"start":3823,"end":3824,"language":96,"track":4400,"speakers":4402,"zh":4417,"en":4420,"tags":4423,"uri":4424},"EHHF78",{"en":167,"zh-hant":10},{"id":3526,"name":4401},{"en":3528,"zh-hant":3528},[4403,4410],{"id":4404,"avatar":4405,"zh":4406,"en":4409},"ZTQ9PZ","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FZTQ9PZ_3TGxCyT.webp",{"name":4407,"bio":4408},"Allen","我目前是台灣科技大學資訊工程所碩士二年級學生，具備生成式 AI、深度學習與 DevOps 的實務經驗。碩士研究聚焦於 LLM 微調的資安議題。專案經驗橫跨大型語言模型等 AI 領域，曾協助跨國企業提出 AI PoC，也參與過 RAG 解決方案的設計。歡迎大家與我交流！",{"name":4407,"bio":4408},{"id":4411,"avatar":4412,"zh":4413,"en":4416},"WHUQPA","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FWHUQPA_DENziWd.webp",{"name":4414,"bio":4415},"黃亮勳","創辦 Twinkle AI，專注於開源繁體中文資料集與下一代繁中模型，為台灣 AI 生態奠定基礎並持續推進模型訓練技術。領導團隊打造台灣首個最小推理模型 Formosa-1 (F1)，並曾以一人團隊完成繁體中文語料蒐集與模型訓練，成功推出台灣首個 Llama 3.2 3B 最小推理模型。",{"name":4414,"bio":4415},{"title":4418,"describe":4419,"type":30},"Benign Outliers 對 Twinkle AI 模型安全對齊之影響研究","你以為「乾淨」的微調資料就是安全的嗎？最新研究指出，僅僅 100 筆來自 Dolly、Alpaca 這類良性資料集的樣本——表面無害、卻帶有有害梯度方向的 Benign Outliers——就足以讓一個對齊過的 LLM 安全防線全面崩潰，而且完全不需要外部惡意資料當錨點。\r\n本場分享將以 Twinkle AI 開源的 gemma-3-4B-T1-it 為實驗對象，帶大家走過一次真實的攻擊與防禦：如何用 Self-Inf-N 從繁中良性資料中挖出這些「隱形地雷」、它們在連續學習場景下有多難消除，以及 SafeGrad、AsFT 這類梯度約束防禦方法到底擋不擋得住。\r\n適合對 LLM 安全、微調、開源模型治理有興趣的開發者與研究者。",{"title":4421,"describe":4422,"type":30},"100 Harmless Samples Are All You Need to Break a Safety-Aligned LLM","Think your \"clean\" fine-tuning data is safe? Recent research shows that just 100 samples from benign datasets like Dolly and Alpaca — innocent on the surface, yet carrying harmful gradient directions — are enough to collapse the safety alignment of an already-aligned LLM. And it works anchor-free: no external malicious data required.\r\nIn this talk, we put Twinkle AI's open-source gemma-3-4B-T1-it on the table for a hands-on attack-and-defense walkthrough: how to surface these \"invisible landmines\" from Traditional Chinese benign datasets using Self-Inf-N, why they're so hard to undo in continual learning scenarios, and whether gradient-constrained defenses like SafeGrad and AsFT can actually hold the line.\r\nFor developers and researchers interested in LLM safety, fine-tuning, and open-source model governance.",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FEHHF78",{"id":4426,"room":4427,"start":4140,"end":3643,"language":42,"track":4428,"speakers":4430,"zh":4438,"en":4441,"tags":4442,"uri":4443},"EXSCKK",{"en":39,"zh-hant":10},{"id":384,"name":4429},{"en":386,"zh-hant":10},[4431],{"id":4432,"avatar":4433,"zh":4434,"en":4437},"UYRMRM","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FUYRMRM_5J2o7Vw.webp",{"name":4435,"bio":4436},"Louis","Ethereum Protocol Developer",{"name":4435,"bio":4436},{"title":4439,"describe":4440,"type":30},"Inside the EVM: From Architecture to Implementation","Developers build DeFi protocols, DAOs, and countless other applications on Ethereum every day using smart contract languages like Solidity and Vyper. Yet beneath all of them sits the Ethereum Virtual Machine (EVM) — the execution engine that powers the entire ecosystem, and still remains a black box to most developers building on it.\r\n\r\nThis talk opens that box from two perspectives.\r\n\r\nFrom the smart contract developer's angle, we'll walk through the EVM's architecture and core components — the stack, memory, transient storage, and more — and show why understanding what happens beneath your Solidity code leads to better gas optimization, safer contracts, and sharper debugging instincts. It's also key to understanding how many smart contract hacks actually work.\r\n\r\nFrom the protocol developer's angle, we'll go behind the scenes of how the EVM itself evolves: how a new feature is proposed, how EIP specs are written, how the implementation comes together, and how multi-client testing keeps consensus intact. We'll give special attention to how gas costs are defined, and the rationale behind the repricing EIPs landing in Ethereum's next hard fork, Glamsterdam.",{"title":4439,"describe":4440,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FEXSCKK",{"id":4445,"room":4446,"start":3969,"end":4447,"language":42,"track":4448,"speakers":4450,"zh":4461,"en":4464,"tags":4465,"uri":4466},"EYEL9B",{"en":684,"zh-hant":10},"2026-08-09T15:20:00+08:00",{"id":686,"name":4449},{"en":688,"zh-hant":689},[4451,4454],{"id":3087,"avatar":3088,"zh":4452,"en":4453},{"name":3090,"bio":3091},{"name":3090,"bio":3091},{"id":4455,"avatar":4456,"zh":4457,"en":4460},"KKMQUS","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FKKMQUS_omRj1YU.webp",{"name":4458,"bio":4459},"Sandra Lin","Sandra 是一名獨立使用者研究員\u002F服務設計顧問，長期參與公民科技與公私協力相關專案，曾貢獻於開源在公共程式平臺上的專案。工作之外，她也是一名斜槓調酒師，跟無數植物的奴隸。",{"name":4458,"bio":4459},{"title":4462,"describe":4463,"type":30},"為什麼開源難以進入標案體系","當數發部推動公共程式，過去唐鳳擔任政委時也曾經推過將開源開放資料精神推入採購範本中，為什麼仍難以看到政府進行開源呢？軟體公司在承接政府標案的過程，為何整體生態仍顯得如此封閉？\r\n\r\n每一個現狀的背後，都有其形成的脈絡。今天的演講，將從兩位講者的視角，拆解開源進入政府體系時可能面臨的阻力：從政府採購制度的運作方式，到實際導入開源的各種挑戰。除了理解這些脈絡，也希望邀請大家一起思考，是否存在能夠施力與介入的空間。",{"title":4462,"describe":4463,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FEYEL9B",{"id":4468,"room":4469,"start":4140,"end":3643,"language":96,"track":4470,"speakers":4472,"zh":4480,"en":4483,"tags":4486,"uri":4487},"F3K9TZ",{"en":3731,"zh-hant":10},{"id":3734,"name":4471},{"en":3736,"zh-hant":3736},[4473],{"id":4474,"avatar":4475,"zh":4476,"en":4479},"EUZF3B","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FEUZF3B_gylCLqw.webp",{"name":4477,"bio":4478},"蕭兆洋（Charles Hsiao）","Site Reliability Engineer，熱愛開源技術，專注於可觀測性、雲端架構與 AIOps。\r\n近期不小心掉進 AI 的深坑裡，時不時會出現在技術社群與大家交流經驗。\r\n深信區塊鏈能如同20世紀末的網際網路般改變世界的發展。\r\n日常活動範圍包含海面下、山林中、或是雲端上。\r\n\r\n[個人部落格: www.charles-hsiao.com](https:\u002F\u002Fwww.charles-hsiao.com)",{"name":4477,"bio":4478},{"title":4481,"describe":4482,"type":30},"邁向 Agentic SRE：以 kagent 與 A2A 協議建構自主化 K8s 維運","隨著 Kubernetes 環境邁向超大規模與多雲複雜化，傳統的 SRE 自動化腳本與預定義規則已難以應對「未知之未知」的系統風險。2026 年的維運思維正在發生移轉——我們不再僅是追求「自動化 (Automation)」，而是致力於實現「自主化 (Autonomy)」。\r\n\r\n本議程將分享如何運用 kagent——CNCF Sandbox 的 Kubernetes 原生 Agentic AI 框架，搭配 A2A (Agent-to-Agent) 協議，讓不同專長的 Agent 能像人類專家一樣相互協商、委派任務並達成共識，在 K8s 上打造一組真正會協作的 SRE Agent。我們會從三個面向切入：\r\n\r\n1. 自主協作架構：A2A 協議如何定義 Agent 間的發現機制、通訊邊界與衝突仲裁，讓異質框架的代理也能無縫協作\r\n2. 從觀測到行動的閉環：kagent 如何讓 Agent 不只「看見」異常，更能協助工程師推理、決策，並在 Kubernetes 與雲端服務上執行修復，與人類團隊共同完成從感知到行動的完整閉環\r\n3. 從意圖到安全執行：將使用者的目標 (Intent) 轉化為可稽核的 Kubernetes 操作，並透過 Human-in-the-Loop 機制守住自動化的邊界",{"title":4484,"describe":4485,"type":30},"Towards Agentic SRE: Building Autonomous K8s Operations with kagent and the A2A Protocol","As Kubernetes environments scale into hyper-scale, multi-cloud complexity, traditional SRE automation scripts and predefined rules struggle to handle the \"unknown unknowns.\" The operational mindset of 2026 is shifting—we're no longer pursuing only Automation, but moving toward Autonomy.\r\n\r\nThis session shares how to leverage kagent—a CNCF Sandbox Kubernetes-native Agentic AI framework—together with the A2A (Agent-to-Agent) Protocol, enabling specialized agents to negotiate, delegate, and reach consensus like human experts, forming a truly collaborative SRE agent team on Kubernetes. We'll explore three dimensions:\r\n\r\n1. Autonomous Collaboration Architecture: How the A2A protocol defines discovery mechanisms, communication boundaries, and conflict arbitration between agents, allowing heterogeneous frameworks to interoperate seamlessly.\r\n2. Closing the Loop from Observation to Action: How kagent empowers agents to not just *detect* anomalies, but to assist engineers in reasoning, decision-making, and executing remediation across Kubernetes and cloud services—completing the full sense-to-act loop alongside human teams.\r\n3. From Intent to Safe Execution: Translating user intent into auditable Kubernetes operations, with Human-in-the-Loop mechanisms safeguarding the boundaries of automation.",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FF3K9TZ",{"id":4489,"room":4490,"start":3709,"end":3823,"language":96,"track":4491,"speakers":4493,"zh":4508,"en":4511,"tags":4512,"uri":4513},"F9VYBK",{"en":3731,"zh-hant":10},{"id":3734,"name":4492},{"en":3736,"zh-hant":3736},[4494,4501],{"id":4495,"avatar":4496,"zh":4497,"en":4500},"TTDMJS","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FTTDMJS_9poyFje.webp",{"name":4498,"bio":4499},"Zespre Chang","Open-source philosopher by conviction, cloud-native engineer by practice. Builds things with Kubernetes so others don't have to reinvent the wheel.",{"name":4498,"bio":4499},{"id":4502,"avatar":4503,"zh":4504,"en":4507},"3EA7BU","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002F3EA7BU_1NbKrJk.webp",{"name":4505,"bio":4506},"Anish Bista","The Kubernetes Guy \r\n\r\nHe is a working professional specializing in DevOps, cloud technologies, Kubernetes, Golang, and the broader cloud-native ecosystem. He is the youngest Kubestronaut from Nepal and an active contributor to the CNCF and open-source ecosystem. He has played a key role in growing the CNCF Kathmandu community and has contributed to projects such as KubeVirt and Kanister. Currently, he is maintaining KubevirtBMC project.",{"name":4505,"bio":4506},{"title":4509,"describe":4510,"type":30},"Bridging the Gap: Bare-Metal Provisioning in a Virtualized World with KubeVirtBMC","Bare-metal provisioning tools speak a language that virtual machines simply don't understand: IPMI and Redfish, the protocols of physical Baseboard Management Controllers (BMCs). Yet in today's cloud-native landscape, there's a very real need to run those same provisioning workflows against virtualized infrastructure: for development, testing, CI pipelines, and beyond.\r\n\r\nIn this talk, we'll introduce KubeVirtBMC, an open-source, cloud-native project that gives KubeVirt VMs virtual BMCs, exposing IPMI and Redfish endpoints so that existing bare-metal provisioning stacks, whether cloud-native or not (Metal3, Tinkerbell, Foreman, and others), can interact with virtual infrastructure without any modification. Think of it as the Kubernetes-native counterpart to VirtualBMC, but built from the ground up around the Kubernetes controller pattern.\r\n\r\nWe'll walk through the architecture, demonstrate a live integration with Metal3\u002FTinkerbell, and discuss project governance, including lessons learned from navigating the migration into the KubeVirt organization.",{"title":4509,"describe":4510,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FF9VYBK",{"id":4515,"room":4516,"start":4281,"end":4282,"language":70,"track":4517,"speakers":4519,"zh":4527,"en":4530,"tags":4531,"uri":4532},"FB3RSE",{"en":93,"zh-hant":10},{"id":3467,"name":4518},{"en":3469,"zh-hant":3469},[4520],{"id":4521,"avatar":4522,"zh":4523,"en":4526},"9VMXTS","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002F9VMXTS_wJgfMXq.jpg",{"name":4524,"bio":4525},"Harunobu Kurokawa","He has been involved in automotive Linux development since 2013 and actively contributes to Linux Foundation projects, including Automotive Grade Linux (AGL). At Renesas, a semiconductor company, he leads embedded Linux development and promotes OSS adoption. Since joining the AGL community in 2016, he has worked on BSP integration, system architecture design, and outreach activities in Japan to support the AGL ecosystem.",{"name":4524,"bio":4525},{"title":4528,"describe":4529,"type":30},"BSP Design for SDVs: Hypervisor and Virtualization Challenges in Decoupling","Software-Defined Vehicle (SDV) requires multiple domains—such as infotainment, instrument clusters, real-time ADAS functions, and AI features —to coexist safely on a single hardware platform. Achieving this requires strong isolation, making virtualization essential. A traditional software coupled with hardware loses software portability.  Hardware decoupling is a crucial aspect in the SDV era.　While multiple approaches exist, this talk focuses on hypervisors, with Xen emerging as a key technology for isolation and hardware decoupling.\r\nHowever, automotive hypervisors have traditionally relied on commercial solutions, and applying an open-source hypervisor like Xen introduces challenges—especially on ARM platforms. These include BSP dependencies, complex boot flows, Device Tree configuration, memory constraints, and virtio-based device design.\r\nIn this talk, we present a real-world hardware implementation combining Xen, Automotive Grade Linux (AGL), and Zephyr RTOS on ARM. We share practical integration issues and solutions, and demonstrate how virtio enables device abstraction and hardware decoupling.",{"title":4528,"describe":4529,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FFB3RSE",{"id":4534,"room":4535,"start":4536,"end":4054,"language":328,"track":4537,"speakers":4539,"zh":4552,"en":4555,"tags":4556,"uri":4557},"FJ7YKS",{"en":522,"zh-hant":10},"2026-08-09T10:50:00+08:00",{"id":3593,"name":4538},{"en":3595,"zh-hant":10},[4540,4547],{"id":4541,"avatar":4542,"zh":4543,"en":4546},"MYUV9K","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FMYUV9K_D53hjaN.webp",{"name":4544,"bio":4545},"Masaya Narita","Founder & Developer, Tac Project\r\nMasaya Narita is an applied chemistry student and the Founder of Tac Project, a student-led indie hardware group in Japan. He explores \"Calm Technology\" to bridge the gap between analog comfort and modern IoT.\r\n\r\nHe is the creator of Tac Photo, a minimalist photo frame powered by FreeRTOS and ESP32. He actively shares his design philosophy and recently published a technical zine about his unique \"screenless\" NFC configuration architecture at a major tech book festival in Japan.",{"name":4544,"bio":4545},{"id":4548,"avatar":3,"zh":4549,"en":4551},"9PZ9EA",{"name":4550,"bio":10},"Kota Mizutani",{"name":4550,"bio":10},{"title":4553,"describe":4554,"type":30},"Tangible Interfaces in Open Hardware: The \"Calm Tech\" Approach to IoT Configuration","Consumer IoT devices, such as digital photo frames, often rely on heavy operating systems (like Android) and touchscreens just to handle basic setups like Wi-Fi and API configurations. This approach not only complicates the hardware design but also limits the device's lifespan—once OS updates stop, the device often becomes obsolete or a security risk.\r\n\r\n\"Tac Photo\" is an open-source digital photo frame project that takes a different approach. By using an ESP32-S3 running FreeRTOS instead of a heavy OS, the device remains lightweight and functional over the long term without requiring frequent software updates.\r\n\r\nTo configure the device without a screen, we separated the setup process from the hardware itself. Users configure their settings (Wi-Fi credentials, cloud sync) on a smartphone app, which encrypts and writes this data to a standard passive NFC card. Tapping this physical card to the frame updates its state instantly.\r\n\r\nIn this session, I will share the architecture and practical lessons learned from building this system. Participants will explore:\r\n\r\n1. Hardware-Software Integration: Using standard NFC to shift complex UI tasks to a smartphone, leaving the hardware simple, single-purposed, and easy to maintain.\r\n\r\n2. Headless Cloud Connections: How to securely transfer encrypted OAuth refresh tokens via NFC to connect to cloud services (like Google Photos) without a device-side browser.\r\n\r\n3. Longevity & Right to Repair: How choosing an RTOS over a full OS, combined with a minimalist, open hardware design, aligns with the Right to Repair movement, empowering users to customize and fix their own digital appliances.\r\n\r\nThis talk provides a practical framework for developers looking to build simple, repairable, and long-lasting IoT devices using FOSS tools.",{"title":4553,"describe":4554,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FFJ7YKS",{"id":4559,"room":4560,"start":3732,"end":3991,"language":70,"track":4561,"speakers":4563,"zh":4571,"en":4574,"tags":4575,"uri":4576},"FNL3K8",{"en":121,"zh-hant":10},{"id":4034,"name":4562},{"en":4036,"zh-hant":4036},[4564],{"id":4565,"avatar":4566,"zh":4567,"en":4570},"37EMMG","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002F37EMMG_c7vLBAn.webp",{"name":4568,"bio":4569},"SAE JIN KIM","I am a student at Woosong University in Daejeon, South Korea. I enjoy building projects driven by my own curiosity and interests, mainly in open source projects, serverless and edge computing.",{"name":4568,"bio":4569},{"title":4572,"describe":4573,"type":30},"Building a Fediverse on the CloudFlare Edge: SiliconBeest","**_SiliconBeest_** is a Fediverse software project designed to run on Cloudflare Workers, an edge-based serverless platform. Inspired by Wildebeest, which has since been deprecated, this project explores how Fediverse services can be built and operated without relying on traditional server infrastructure.\r\n\r\nIn this talk, I will share why I chose Cloudflare Workers as the foundation for building Fediverse software, how SiliconBeest is being implemented, and what this architecture could mean for the future of federated social platforms. Drawing from ideas and experience from Wildebeest, Mastodon, Misskey, and Fedify, I am building a fully functional Fediverse software stack that aims to be practical, scalable, and affordable.\r\n\r\nOne of the key ideas behind this project is that a serverless edge architecture can reduce operational complexity while making it easier to handle growth. Instead of depending on large dedicated servers, the system is designed to scale with demand. In the talk, I will discuss the technical design, the advantages and trade-offs of running Fediverse software in this environment, and what kinds of scale may be achievable starting from a relatively low cost, such as Cloudflare Workers’ paid plan.\r\n\r\nI will also talk about the development process behind the project. AI coding tools such as Claude Code and Codex have been a meaningful part of how I design, implement, and iterate on SiliconBeest. I want to share how these tools have helped in practice, where they were useful, where they were not, and what it feels like to build an ambitious open software project with AI as part of the workflow.\r\n\r\nThis session is for people interested in building Fediverse software that is scalable, cost-effective, and easy to maintain, and in exploring how serverless can work with the Fediverse.",{"title":4572,"describe":4573,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FFNL3K8",{"id":4578,"room":4579,"start":4536,"end":4054,"language":42,"track":4580,"speakers":4582,"zh":4590,"en":4593,"tags":4594,"uri":4595},"FSL3ZL",{"en":39,"zh-hant":10},{"id":384,"name":4581},{"en":386,"zh-hant":10},[4583],{"id":4584,"avatar":4585,"zh":4586,"en":4589},"F378KS","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002F8223657_Qepvwga.png",{"name":4587,"bio":4588},"Kevin Chia","Previously a software engineer in blockchain industry, now independently learning zk and lattice-based cryptography.",{"name":4587,"bio":4588},{"title":4591,"describe":4592,"type":30},"Intro to folding schemes and its lattice variant","This talk will give an introduction to folding schemes which are widely used to reduce many incremental verifiable computations to a smaller one in zk proving systems. Original folding schemes  (Nova, HyperNova) are based on elliptic curve cryptography, whose security is based on discrete log problem and is subject to quantum computer attacks. \r\n\r\nTo address this, new schemes have been designed using post quantum cryptography, and lattice-based is one of them. The audience will then learn the intuitions and techniques behind these lattice-based folding schemes, as well as the challenges they face.",{"title":4591,"describe":4592,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FFSL3ZL",{"id":4597,"room":4598,"start":3505,"end":3991,"language":328,"track":4599,"speakers":4601,"zh":4609,"en":4612,"tags":4613,"uri":4614},"FXNFZM",{"en":922,"zh-hant":10},{"id":330,"name":4600},{"en":332,"zh-hant":332},[4602],{"id":4603,"avatar":4604,"zh":4605,"en":4608},"9NDCKB","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002F9NDCKB_iwcaFmL.svg",{"name":4606,"bio":4607},"Laksh Gambhir","Laksh Gambhir is a builder at heart who enjoys getting unreasonably deep into systems, figuring out how they work, where they break, and how to make them smoother.\r\nHe works on electronic system-level modelling, virtual prototyping, and low-level tooling, often around Linux workflows and automation. Having worked in visual design as well, he tends to obsess over making systems cleaner, simpler, and nicer to use.",{"name":4606,"bio":4607},{"title":4610,"describe":4611,"type":59},"Controlling Ubuntu with Local LLM: A Practical System Using Structured Tool Interfaces","Modern Ubuntu systems offer powerful tools like systemd, LXD, and various CLI utilities, but these often remain fragmented and require significant command-line expertise to operate efficiently. In this talk, we will present a system that enables natural language control of Ubuntu using a fully local AI Model, our system contains:\r\n- A local LLM for intent interpretation\r\n- A lightweight orchestration layer\r\n- Tool adapters connected to Ubuntu components (systemd, LXD, logs)\r\n\r\nInstead of relying on cloud services, this system uses a local LLM combined with a structured tool interface to safely and predictably interact with system components. This allows users to monitor performance, manage services, control LXD containers, and summarize logs using conversational commands.\r\n\r\nMore info: https:\u002F\u002Fevents.canonical.com\u002Fevent\u002F146\u002Fcontributions\u002F954\u002F",{"title":4610,"describe":4611,"type":59},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FFXNFZM",{"id":4616,"room":4617,"start":3568,"end":3548,"language":42,"track":4618,"speakers":4620,"zh":4628,"en":4631,"tags":4634,"uri":4635},"FYSGQ8",{"en":237,"zh-hant":10},{"id":241,"name":4619},{"en":243,"zh-hant":243},[4621],{"id":4622,"avatar":4623,"zh":4624,"en":4627},"E7EMQK","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FE7EMQK_LXgQauG.jpeg",{"name":4625,"bio":4626},"Kasa","[Kasa](https:\u002F\u002Fx.com\u002Fk_hno3) is a web developer juggling Ruby, TypeScript, Python, and a bit of DevOps. She has been one of the core organizers of the [Ruby Taiwan Community ](https:\u002F\u002Flinktr.ee\u002Frubytaiwan2008)since August 2025.\r\n\r\nShe volunteered at [RubyKaigi 2026](https:\u002F\u002Frubykaigi.org\u002F2026\u002Fschedule\u002Fday1\u002F) and is currently recovering from post-Hakodate maguro withdrawal.\r\n\r\nHer favorite Ruby syntax is inheritance: `class Kasa \u003C Hsiao`. Something about its conciseness and quiet wit soothes her inner chaos.",{"name":4625,"bio":4626},{"title":4629,"describe":4630,"type":30},"聊聊 Ruby 的 ReDoS 防禦：吃瓜設計權衡、整合 Rails 實踐、窺探 Regexp 未來動向","「每次提起升版就被已讀。」\r\n\r\nRuby 3.2 導入了重要的 Regexp 更新，而隨著 Ruby 4 發佈，Ruby 3.2 也正式進入 EOL。近年的 Ruby Regexp 改善，也讓 ReDoS 的防禦逐漸從「怎麼寫 Regex」延伸到「底層如何幫開發者收斂風險」。\r\n\r\n然而， 2026 年初 GitLab 的 CVE-2026-1388，以及 Active Support 的 CVE-2026-33169，也再次提醒我們：在複雜的 Web 應用實作中，Regexp 仍然可能成為效能與安全問題的入口。\r\n\r\n這場演講會延續 [RubyJam 3 月 Meetup](https:\u002F\u002Frubytaiwan.kktix.cc\u002Fevents\u002Frubyjam2603) 中對 Ruby Feature [#17837](https:\u002F\u002Fbugs.ruby-lang.org\u002Fissues\u002F17837) 與 [#19104](https:\u002F\u002Fbugs.ruby-lang.org\u002Fissues\u002F19104) 的設計權衡討論，並加入 2026 年最新的技術動態：\r\n\r\n1. 案例拆解：分析 2026 年兩起 ReDoS CVE，看看在現有防護機制下，哪些模式依然可能帶來風險。\r\n2. Rails 實踐：說明如何在 Rails 應用程式中整合 Regexp Timeout。\r\n3. 從 ReDoS 延伸到 Timing Attack：回應 RubyJam Meetup 的現場提問，聊聊演算法複雜度攻擊的另一面，以及 Ruby 生態系中如何實作 Constant Time Comparison。\r\n4. Ruby Regexp 的最新方向：分享來自 [RubyKaigi 2026 ](https:\u002F\u002Frubykaigi.org\u002F2026\u002F)的第一手資訊，看看 Regexp 導入 JIT 的進展，以及對安全性議題的可能影響。",{"title":4632,"describe":4633,"type":30},"Diving into Ruby 3 ReDoS Defense: Design Trade-offs, Rails Integration, and the Future of Regexp","*\"Left on read... every time I suggest a version upgrade.\"*\r\n\r\nRuby 3.2 introduced significant improvements to Regexp security and performance. With Ruby 4 recently released and Ruby 3.2 now officially reaching EOL, the community’s approach to ReDoS (Regular Expression Denial of Service) has gradually extended from “how do we write safe Regex” to “how can the runtime itself help reduce risk for developers.”\r\n\r\nHowever, the CVE-2026-1388 incident in GitLab and CVE-2026-33169 in Active Support in early 2026 once again remind us that in real-world web applications, Regexp can still become an unexpected source of performance and security issues.\r\n\r\nThis talk continues the discussion from the RubyJam 2026.03 Meetup around Ruby Feature #17837 and #19104, and extends it with the latest developments in 2026:\r\n\r\n1. Case breakdown: Analyzing two ReDoS-related CVEs in 2026, and examining which Regexp patterns can still introduce risk under existing mitigation mechanisms.\r\n2. Rails in practice: How to integrate Regexp Timeout into Rails applications, and what practical pitfalls to watch out for.\r\n3. From ReDoS to Timing Attacks: Following up on questions from the RubyJam Meetup, exploring the other side of algorithmic complexity attacks, and how Ruby ecosystems implement Constant Time Comparison.\r\n4. The latest direction of Ruby Regexp: First-hand insights from RubyKaigi 2026, looking into the ongoing progress of introducing JIT into Regexp and what it may imply for security considerations going forward.",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FFYSGQ8",{"id":4637,"room":4638,"start":4639,"end":4640,"language":96,"track":4641,"speakers":4643,"zh":4651,"en":4654,"tags":4655,"uri":4656},"G7EXXK",{"en":407,"zh-hant":10},"2026-08-09T10:05:00+08:00","2026-08-09T10:35:00+08:00",{"id":410,"name":4642},{"en":412,"zh-hant":412},[4644],{"id":4645,"avatar":4646,"zh":4647,"en":4650},"ZNLX7K","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FZNLX7K_YfDDxGu.png",{"name":4648,"bio":4649},"小米","不務正業的後端工程師\r\n嘗試用程式把腦中的世界呈現給其他人",{"name":4648,"bio":4649},{"title":4652,"describe":4653,"type":30},"前端工程師也能做互動藝術：用 JavaScript、Canvas 與 TouchDesigner 打開創作的大門","過去提到 TouchDesigner，很多前端工程師可能會覺得那是屬於互動藝術家、VJ、裝置藝術或 Python 使用者的工具；但如果我們換個角度，把 TouchDesigner 看成一個可以收發 HTTP \u002F WebSocket、也能渲染網頁內容的即時視覺引擎，它其實可以和 JavaScript 生態系自然地接在一起。\r\n\r\n這場演講會分享如何用前端工程師熟悉的技術——Canvas、p5.js、three.js、HTTP API 與瀏覽器——來控制 TouchDesigner，並把它應用在互動動畫、即時多人遊戲與 AI 輔助生成視覺效果上。你不需要先成為專業互動藝術家，也不需要從零開始學完整的 TouchDesigner 節點系統；只要你會寫 JavaScript，就可以把既有的 Web 技術帶進即時影像與互動藝術的世界。",{"title":4652,"describe":4653,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FG7EXXK",{"id":4658,"room":4659,"start":3824,"end":3590,"language":96,"track":4660,"speakers":4662,"zh":4670,"en":4673,"tags":4674,"uri":4675},"G8QTLU",{"en":167,"zh-hant":10},{"id":3526,"name":4661},{"en":3528,"zh-hant":3528},[4663],{"id":4664,"avatar":4665,"zh":4666,"en":4669},"78YLDA","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002F78YLDA_AtWhJQc.webp",{"name":4667,"bio":4668},"老鮑伯","本人於 2007 年底投身資訊業，從原本寫 Java， 2009 年開始接觸 WPF、Silverligh t的 C# 桌面應用程式開始投入 .NET開發，之後於資拓宏宇擔任軟體專案工程師，於 2014 年投入遊戲業在台灣卡普空擔任開發組組長，2018 到2023 上半在XR領域相關新創擔任 CTO； .NET 相關的前後端與Azure雲端開發技術都略懂略懂，並在近期積極往 Agentic AI等相關AI應用方向鑽研，但抓 bug 比較在行。",{"name":4667,"bio":4668},{"title":4671,"describe":4672,"type":30},"在地的模型當然要可以跑在地端","介紹如何以 GPU 和 NPU 在地端跑 LLM 模型推論運算的運作機制，以及介紹如何用幾種開源\u002F非開源的框架([Ollama](https:\u002F\u002Follama.com)\u002F[vLLM](https:\u002F\u002Fvllm.ai\u002F)\u002F[LM Studio](https:\u002F\u002Flmstudio.ai\u002F)\u002F[Lemonade](https:\u002F\u002Flemonade-server.ai\u002F)\u002F[oBeaver](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002Fobeaver))在地端不同種 GPU\u002FNPU 的機器上執行 Twinkle AI 開源模型。",{"title":4671,"describe":4672,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FG8QTLU",{"id":4677,"room":4678,"start":3464,"end":3465,"language":96,"track":4679,"speakers":4681,"zh":4689,"en":4692,"tags":4695,"uri":4696},"GEKJRV",{"en":263,"zh-hant":10},{"id":44,"name":4680},{"en":46,"zh-hant":47},[4682],{"id":4683,"avatar":4684,"zh":4685,"en":4688},"UAREZS","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FUAREZS_aAHZz5N.webp",{"name":4686,"bio":4687},"Kuan-Ju Chou","在台灣人權促進會服務，負責數位人權。",{"name":4686,"bio":4687},{"title":4690,"describe":4691,"type":30},"健保資料庫案之後：停止利用權如何實踐？以及其他大型資料庫","醫療隱私，你在意嗎？本場將會簡介如何申請停止利用權、目前的限制，還有台灣現存具有大量實名個資的政府資料庫。\r\n\r\n台灣的就醫資料多年來，未經病人同意，分享給產業與學界使用。自台灣人權促進會、台灣女人連線、健保監督聯盟三個民間團體發動退出權訴訟以來，健保資料庫目的外利用陸續停止販售光碟、限縮至僅供研究使用。直到2022年憲法訴訟最終確認政府必須讓人民實踐停止利用權，也就是不同意將個資提供其他目的使用。去年，台灣終於完成修法，開始讓民眾可以行使停止利用權，但在立法過程中卻增加不少例外，大幅縮小停止利用權的範圍。\r\n\r\n除此之外，台灣還有不少法律規範不明、對個資當事人權利保障尚不清楚的資料庫，需要公民社會一起監督。",{"title":4693,"describe":4694,"type":30},"How to opt-out secondary use of National Health Insurance Data? Massive Databases you might not Aware","Medical Privacy: Do You Care?\r\n\r\nAn introduction to opting out of data sharing, current limitations, and examples of  the landscape of government databases in Taiwan.\r\n\r\nDo you care about your medical privacy? For years, national health insurance data in Taiwan have been shared with industry and academia without consent. It is mandatory insurance, which means almost all the population in Taiwan have joined the system.\r\n\r\nFollowing a series of \"Right to Opt-Out\" lawsuits spearheaded by three NGOs—the Taiwan Association for Human Rights (TAHR), Taiwan Women's Link, and the NHI Supervision Alliance—the secondary use of National Health Insurance (NHI) data has gradually shifted. Changes included halting the sale of data on physical discs and restricting access strictly to research purposes.\r\nIn 2022, a landmark Constitutional Court ruling finally confirmed that the government must allow citizens to exercise their Right to Object (the right to stop the use of their data), ensuring individuals can refuse to have their personal information used for purposes other than their original intent. While Taiwan finally amended its laws last year to enable this right, several exceptions were added during the legislative process, significantly narrowing the actual scope of the \"Right to Opt-Out.\"\r\n\r\nWhy This Matters\r\n\r\nBeyond the NHI database, Taiwan maintains numerous other databases containing vast amounts of identifiable personal data. Many of these operate under vague legal frameworks with unclear protections for data subjects' rights. Now more than ever, these systems require vigilant oversight from civil society.",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FGEKJRV",{"id":4698,"room":4699,"start":3909,"end":4700,"language":96,"track":4701,"speakers":4703,"zh":4707,"en":4710,"tags":4711,"uri":4712},"GGXEWB",{"en":684,"zh-hant":10},"2026-08-09T13:50:00+08:00",{"id":686,"name":4702},{"en":688,"zh-hant":689},[4704],{"id":2156,"avatar":2157,"zh":4705,"en":4706},{"name":2159,"bio":2160},{"name":2159,"bio":2160},{"title":4708,"describe":4709,"type":59},"政府採購如何使用開源？為何開源","政府資訊服務導入開源，一方面有助於降低廠商鎖定風險，減少機關長期採購與維護成本；另一方面，也能進一步讓維護者、其他機關與外部協作者更容易理解、延續並改善既有服務。不過這兩項優勢，必須分階段看待，前者比後者務實且急迫。\r\n\r\n然而，開源仍未成為政府資訊採購的主流實務。從需求規格、招標文件、驗收標準到後續維運，行政流程中仍有許多尚待調整的環節。本場對談將從台灣與美國地方政府的開源採購經驗出發，討論開源在政府資訊服務中的實際應用、已經取得進展的制度條件，以及未來或可重新設計的採購與行政流程。",{"title":4708,"describe":4709,"type":59},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FGGXEWB",{"id":4714,"room":4715,"start":3709,"end":3823,"language":96,"track":4716,"speakers":4718,"zh":4726,"en":4729,"tags":4730,"uri":4731},"GLFDKK",{"en":407,"zh-hant":10},{"id":410,"name":4717},{"en":412,"zh-hant":412},[4719],{"id":4720,"avatar":4721,"zh":4722,"en":4725},"HM3LDG","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FHM3LDG_ZmSfRxJ.jpg",{"name":4723,"bio":4724},"gholk","linux user, interested driven web developer, WebExtension developer.\r\nmastodon: https:\u002F\u002Ftaiwan.wtf\u002F@gholk",{"name":4723,"bio":4724},{"title":4727,"describe":4728,"type":30},"Tridactyl - 像個 vim 高手與 JS 開發者一樣使用 Firefox","Tridactyl 是能全鍵盤操作瀏覽器的 Firefox 擴充元件，類似於 Vimium，但他的能力遠不只於此。\r\n\r\nTridactyl 有任意執行 js 、呼叫 WebExtensions API 的能力，核心的 vim 風格操作介面、 ex 指令列，搭配上極為開放的 API，讓 Tridactyl 幾乎可以取代大多數的擴充元件、油猴腳本、 bookmarklet。\r\n\r\nTridactyl 更是一款讓 js 開發者能完全掌握瀏覽器的擴充元件，有了他，不再需要在開發者工具裡貼貼剪剪程式碼片段，原本只是臨時寫就的程式碼，可以直接記憶到 vim 風格的快捷鍵組合，或是單純儲存為 ex 命令。tridactyl 讓 js 開發者能將自己對程式語言的理解，實際融入瀏覽器的日常使用當中；此時正是 JS 開發者賦權的時刻！",{"title":4727,"describe":4728,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FGLFDKK",{"id":4733,"room":4734,"start":3504,"end":4735,"language":4736,"track":4737,"speakers":4739,"zh":4759,"en":4762,"tags":4765,"uri":4766},"GLX9HZ",{"en":1007,"zh-hant":10},"2026-08-09T12:30:00+08:00","Others",{"id":901,"name":4738},{"en":903,"zh-hant":904},[4740,4747,4752],{"id":4741,"avatar":4742,"zh":4743,"en":4746},"LEFUDE","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FLEFUDE_MLhsYO0.png",{"name":4744,"bio":4745},"John Ho 何重義","John Ho is a passionate mindfulness advocate and tech veteran with over 25 years of experience in customer experience (CX) leadership at companies like Microsoft and VMware. For the past 5+ years, he has shared insights from his personal mindfulness ritual practice through weekly meditative walk posts on LinkedIn, inspiring a global audience.  John blends CX wisdoms with learnings from his urban nature walks to help individuals reconnect with their authentic selves—especially vital in today’s fast-moving, AI-driven world.  John is also inspired to build a global mindfulness community where people can come together and exchange mindfulness best practices and resources. https:\u002F\u002Fwww.linkedin.com\u002Fin\u002Fjohnho\u002F\r\n何重義 是一位充滿熱情的正念倡導者，同時也是一位技術資深人士，在微軟和 VMware 等公司擁有超過 25 年的客戶體驗 (CX) 領導經驗。過去五年多來，他透過每週在領英 (LinkedIn) 上發布冥想散步帖子，分享個人正念儀式實踐的感悟，激勵著全球受眾。 John 將客戶體驗智慧與城市自然漫步的學習成果相結合，幫助人們重新找回真實的自我——這在當今快速發展的人工智慧驅動型世界中尤其重要。 John 也致力於建立一個全球正念社區，讓人們聚集在一起，交流正念的最佳實踐和資源。",{"name":4744,"bio":4745},{"id":4748,"avatar":3,"zh":4749,"en":4751},"GHRKTH",{"name":4750,"bio":10},"Jeffrey",{"name":4750,"bio":10},{"id":4753,"avatar":4754,"zh":4755,"en":4758},"REB9UB","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FREB9UB_4ya4gfT.jpeg",{"name":4756,"bio":4757},"徐一瑄（Stephy Hsu）","**學歷**\r\n阿德勒大學臨床心理學博士在讀 Adler University, Psy.D Clinical Psychology\r\n哥倫比亞大學臨床心理學碩士 TC Columbia University, MA Clinical Psychology\r\n紐約大學藝術治療碩士 New York University, MA Art Therapy\r\n帕森斯藝術學院純藝術、心理學和商 Parsons The New School, BFA Fine Arts, Psychology, Business\r\n**經歷**\r\n美國紐約藝術治療學會公關主席 New York Art Therapy Association Public Relations Chair\r\n美國紐約賴克斯島監獄重症精神病房 (Riker's Island Correctional Facility ) 實習藝術治療師\r\n美國紐約兒童中心 (The Child Center of NY), 4-10歲幼童、青少年自傾向群體、產前\u002F後憂鬱症、家庭\u002F伴侶藝術治療師\r\n美國紐約西奈山伊坎醫學院 (Icahn School of Medicine at Mount Sinai) 自殺預防研究部門研究人員\r\n美國哈佛醫學院附屬布萊根婦女醫院 (Brigham and Women’s Hospital) 研究人員\r\n加拿大 BC Bereavement Helpline 哀傷支持與社區服務實習生",{"name":4756,"bio":4757},{"title":4760,"describe":4761,"type":59},"（中英雙語）走入正念: 職場及社群參與身心靈平衡經驗分享","歡迎參加我們為沉浸式的正念體驗活動，一同探索以儀式培養內在平靜與自我覺知。活動由John Ho 主持，他將分享他個人和家人正念之旅的感悟。本地的身心靈合作夥伴: Bernice 蔡碧瑜, Newfound Life《怒放生活》創辦人｜個人成長｜健康管理師; Jessie Chang 聊心茶室 (HealYou); Jeffrey Schwab\r\n, Chief Possibility Officer of Uncertain Pathfinders-Ex CLO-APTD-Lifelong improviser-Lumina Spark Practitioner-Coach\u002FMentor-Writer-Process Oriented Extrovert-Banjo Player-Embracer of Trial and Error;  Stephy 徐一瑄, 美國註冊藝術治療師 (ATR-BC), 紐約州創意藝術治療師 (LCAT); 也將介紹她們台灣正念領域的資源。您將透過與導師及與會者的分組互動，深度的探索最適合您的正面實踐方式。\r\n \r\n本活動將中英雙語進行。\r\n\r\nJoin us for an immersive 3-hour mindfulness experience where we’ll explore the power of ritual in cultivating inner calm and self-awareness. Host John Ho will share insights from his personal and family mindfulness journeys combined with his years of professional mind connection learnings. Local wellness partners: Bernice 蔡碧瑜, Newfound Life《怒放生活》創辦人｜個人成長｜健康管理師; Jessie Chang 聊心茶室 (HealYou); Jeffrey Schwab, Chief Possibility Officer of Uncertain Pathfinders-Ex CLO-APTD-Lifelong improviser-Lumina Spark Practitioner-Coach\u002FMentor-Writer-Process Oriented Extrovert-Banjo Player-Embracer of Trial and Error;  Stephy 徐一瑄, 美國註冊藝術治療師 (ATR-BC), 紐約州創意藝術治療師 (LCAT) will also introduce their resources unique to Taiwan’s mindfulness scene. Through engaging breakout sessions, you’ll have the opportunity to connect deeply with facilitators and fellow participants to discover practices that resonate with you.",{"title":4763,"describe":4764,"type":59},"Mindful Connection Collective","Join us for an immersive 3-hour mindfulness experience where we’ll explore the power of ritual in cultivating inner calm and self-awareness. Host John Ho will share insights from his personal and family mindfulness journeys combined with his years of professional mind connection learnings. Local wellness partners: Bernice 蔡碧瑜, Newfound Life《怒放生活》創辦人｜個人成長｜健康管理師; Jessie Chang 聊心茶室 (HealYou); Jeffrey Schwab, Chief Possibility Officer of Uncertain Pathfinders-Ex CLO-APTD-Lifelong improviser-Lumina Spark Practitioner-Coach\u002FMentor-Writer-Process Oriented Extrovert-Banjo Player-Embracer of Trial and Error;  Stephy 徐一瑄, 美國註冊藝術治療師 (ATR-BC), 紐約州創意藝術治療師 (LCAT) will also introduce their resources unique to Taiwan’s mindfulness scene. Through engaging breakout sessions, you’ll have the opportunity to connect deeply with facilitators and fellow participants to discover practices that resonate with you.",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FGLX9HZ",{"id":4768,"room":4769,"start":3567,"end":3568,"language":70,"track":4770,"speakers":4772,"zh":4780,"en":4783,"tags":4784,"uri":4785},"GQNR79",{"en":121,"zh-hant":10},{"id":4034,"name":4771},{"en":4036,"zh-hant":4036},[4773],{"id":4774,"avatar":4775,"zh":4776,"en":4779},"HVGEET","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FHVGEET_aPDRsba.webp",{"name":4777,"bio":4778},"Haze","Haze is a software engineer interested in open source projects, the Fediverse, and developer tools. He created ActivityPlug and mainly contributes to the iOS client for HackersPub, a Fediverse-based developer community. His professional work focuses on software development in the healthcare and medical domain. He is also involved in awareness activities in South Korea for migraine and other headache disorders.",{"name":4777,"bio":4778},{"title":4781,"describe":4782,"type":30},"ActivityPlug: a unified API layer for ActivityPub server software","ActivityPub servers can communicate with each other through the ActivityPub protocol, but their application APIs are not uniform. A client, bot, or integration that works with one server implementation often needs separate adapters, authentication handling, data mapping, and error handling to support another.\r\n\r\nThis talk introduces ActivityPlug, a library and proxy API server that wraps different ActivityPub server APIs behind one consistent interface. The project is designed for developers who want to build tools for the Fediverse without binding their code to a single server implementation.\r\n\r\nActivityPlug can be used in two modes. As a proxy API server, it exposes GraphQL and HTTP APIs that applications can call from any runtime environment. In the JavaScript ecosystem, it can also be used directly as a library, without running a separate server. This makes it suitable for web clients, mobile clients built with technologies such as React Native, bots, automation tools, and connected services that operate across ActivityPub servers.\r\n\r\nThe presentation will cover the design problem, the current architecture, and examples of how ActivityPlug normalises server-specific APIs into one unified interface. It will also show how a shared API layer can reduce the cost of building tools for the Fediverse: a client or service can target ActivityPlug once, then support multiple ActivityPub server implementations through adapters.\r\n\r\nThe goal is not to replace existing ActivityPub servers or define a new social protocol. ActivityPlug is a practical compatibility layer for application developers. By making server APIs easier to work with, it aims to help more people build clients, bots, bridges, moderation tools, analytics tools, and other services on top of the Fediverse.",{"title":4781,"describe":4782,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FGQNR79",{"id":4787,"room":4788,"start":3504,"end":3505,"language":96,"track":4789,"speakers":4791,"zh":4799,"en":4802,"tags":4805,"uri":4806},"GTQ3WH",{"en":9,"zh-hant":10},{"id":3645,"name":4790},{"en":3647,"zh-hant":3648},[4792],{"id":4793,"avatar":4794,"zh":4795,"en":4798},"PEGEJN","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002F29%E6%AD%B2%E5%A5%B3%E7%94%9FFB%E9%80%A3%E7%B5%90%E5%9C%96_rcupdt3.jpg",{"name":4796,"bio":4797},"TZU","FB:29歲女生學程式的小編",{"name":4796,"bio":4797},{"title":4800,"describe":4801,"type":30},"Laravel 實戰：用 hash 比對實作以圖搜圖","本議程將以 Laravel 實戰方式，示範如何用「圖片雜湊（image hash）比對」實作以圖搜圖功能。內容包含：在 Laravel 中處理圖片上傳與儲存、使用感知雜湊（aHash \u002F pHash）計算圖片特徵、將 hash 值存入資料庫，以及透過漢明距離比對找出相似圖片。最後會說明這種做法的優缺點，以及如何擴展成更完整的以圖搜圖系統，適合熟悉 Laravel 的 PHP 開發者學習實務應用。",{"title":4803,"describe":4804,"type":30},"Laravel in Practice: Implementing Image‑by‑Image Search with Hash Comparison","This session will demonstrate, in a hands‑on Laravel style, how to implement image‑by‑image search using “image hash comparison.” The content covers handling image upload and storage in Laravel, using perceptual hashing (aHash \u002F pHash) to compute image features, storing hash values in the database, and finding similar images by calculating Hamming distance. The talk will also discuss the advantages and limitations of this approach, and how it can be extended into a more complete image‑search system, making it suitable for PHP developers familiar with Laravel who want to learn practical real‑world usage.",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FGTQ3WH",{"id":4808,"room":4809,"start":3865,"end":4810,"language":96,"track":4811,"speakers":4813,"zh":4821,"en":4824,"tags":4825,"uri":4826},"H3VJLX",{"en":326,"zh-hant":10},"2026-08-09T15:55:00+08:00",{"id":3268,"name":4812},{"en":3270,"zh-hant":3271},[4814],{"id":4815,"avatar":4816,"zh":4817,"en":4820},"BMWNNY","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FBMWNNY_xVbNEKG.webp",{"name":4818,"bio":4819},"Jarvis Chen","曾任軟體開發工程師，目前擔任 DevOps 工程師，專注於系統現代化與自動化部署相關工作。身為一位熱衷研究新技術的工程師，我目前的重心在於 CNCF 生態系中的雲原生技術實務落地，以及 AI Agent 的開發與創新應用。\r\n\r\n我非常享受深入研究最新技術趨勢，並將這些經驗轉化為內部技術分享，協助團隊建立更成熟的 DevOps 文化。我也期待在快速變化的技術環境中，持續迎接新的挑戰與成長。",{"name":4818,"bio":4819},{"title":4822,"describe":4823,"type":30},"打開就能工作的 AI Agent：Aileron 的 Workspace 設計實踐","AI Agent 已經不只是一個會寫程式的工具。從 Claude Code、Codex、Gemini 到OpenCode，不同 Agent 能力正在快速演進；但對企業來說，真正困難的往往不是模型夠不夠強，而是如何讓 Agent 在符合內部規範、權限治理與基礎設施要求的前提下，被團隊穩定採用。\r\n\r\n實際導入時，使用者常常還是得自己處理 system prompt、MCP、skills、slash commands、工具安裝、SDK、CLI、專案起始檔案與執行環境。這些設定散落在個人經驗、團隊文件與本機環境裡，讓 AI Agent 很難從工程師個人工具，變成企業可管理、可複製、可擴展的工作方式。\r\n\r\n本次將介紹 Aileron 開源平台, Aileron 的目標不是只讓 AI 更會寫 code，而是讓組織可以用標準化的方式提供 Agent 工作區，讓使用者打開 workspace 就能開始工作。\r\n\r\n  為了讓這件事成立，Aileron 聚焦在三個關鍵設計：\r\n\r\n  1. 讓 Agent 更容易被企業採用\r\n     透過集中式 Marketplace、模板、權限與標準化能力，將 system prompt、MCP、skills、slash commands、起始檔案與工具需求包裝成可分發、可選擇、可治理的工作區能力。企業不需要依賴每個人手動拼裝 Agent 設定，而是可以用平台化方式管理不同場景的 Agent 使用方式。\r\n\r\n  2. 讓使用者更容易開始工作\r\n     Aileron 將 Chat、檔案管理、Git、Web Terminal、OpenSpec 與常用 CLI \u002F SDK 整合在同一個 workspace 中，並透過 base image 預先準備執行環境。使用者不需要先安裝一堆工具、處理權限或設定本機環境，就能直接進入可工作的Agent workspace。\r\n\r\n  3. 與 Kubernetes 結合，支撐正式環境部署\r\n     Aileron 透過容器化 runtime 與 workspace-operator，將每個 Agent workspace 轉換成 Kubernetes 中可動態佈建、隔離、重現與管理的執行環境。這讓 Agent 不只是換 prompt，而是真的進入對應任務的工作場景，也讓企業能用既有的 K8s來承載 Agent 工作負載。\r\n\r\n  本次分享重點包括：\r\n  1. 導入 AI Agent 時，從個人工具走向平台化工作區會遇到哪些門檻\r\n  2. Aileron 如何用 Marketplace \u002F 模板系統，讓 Agent 能力變成可分發與重用的 workspace capability\r\n  3. base image、容器化與預裝工具，如何降低使用者開始工作的成本\r\n  4. workspace-operator 如何結合 Kubernetes，動態建立可隔離、可重現、可管理的 Agent workspace\r\n  5. 使用 OpenSpec 開發 Aileron 的歷程與實務經驗",{"title":4822,"describe":4823,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FH3VJLX",{"id":4828,"room":4829,"start":4830,"end":4831,"language":328,"track":4832,"speakers":4834,"zh":4842,"en":4845,"tags":4846,"uri":4847},"H7JATM",{"en":121,"zh-hant":10},"2026-08-09T15:30:00+08:00","2026-08-09T16:00:00+08:00",{"id":4034,"name":4833},{"en":4036,"zh-hant":4036},[4835],{"id":4836,"avatar":4837,"zh":4838,"en":4841},"DNXZAB","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FDNXZAB_4Tcagsw.webp",{"name":4839,"bio":4840},"Jiajun Xu","- Drupal site builder and third-party developer for Nextcloud\r\n- Board member of [Software Liberty Association Taiwan (SLAT)](https:\u002F\u002Fslat.org.tw)\r\n- Author of *Software Liberty Newsletter*, with a long-term focus on the Fediverse across Asia; last year, [interviewed](https:\u002F\u002Fblogs.slat.org.tw\u002Fnode\u002F1031) Sandy Corzeta, administrator of [misskey.id](https:\u002F\u002Fmisskey.id)\r\n- Sociologist by training, with previous research on migrant workers in Asia\r\n- Living in a Taiwanese-Indonesian household, with active ties to the Indonesian FLOSS community",{"name":4839,"bio":4840},{"title":4843,"describe":4844,"type":30},"From CMS to Fediverse: Drupal's ActivityPub module","What if your CMS could not only host a collaborative blog or a news site, but also directly become a node on the Fediverse — followable, replyable, boostable by Mastodon users?\r\n\r\nThe Drupal [ActivityPub](https:\u002F\u002Fwww.drupal.org\u002Fproject\u002Factivitypub) module is turning this into a \"just build a website\" task. Leveraging Drupal's long-standing strength in flexible content modeling, any content type can map to an ActivityStreams object type, every user is an actor, and pairing it with the Blog module gives you a multi-author publishing platform that the Fediverse can follow.\r\n\r\n[Funded](https:\u002F\u002Frealize.be\u002Fnotes\u002F2327) by NLnet, maintainer [Swentel](https:\u002F\u002Fwww.drupal.org\u002Fu\u002Fswentel) is sprinting toward the [1.0.0](https:\u002F\u002Fwww.drupal.org\u002Fproject\u002Factivitypub\u002Fissues\u002F3572274) release. Major milestones including Shared Inbox and the Mastodon-compatible API have already landed.\r\n\r\nMeanwhile, last year's release of the [Drupal CMS](https:\u002F\u002Fnew.drupal.org\u002Fdrupal-cms) distribution and features like [Recipes](https:\u002F\u002Fwww.drupal.org\u002Fdocs\u002Fextending-drupal\u002Fdrupal-recipes) have made Drupal genuinely install-and-go. The pieces are falling into place — this path is finally accessible to non-developers.\r\n\r\nIn this talk I'll walk through how the module works, demo a multi-author blog being followed from Mastodon live on stage, and share what I've observed as a site builder and small-time contributor about what Drupal brings to the Fediverse.",{"title":4843,"describe":4844,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FH7JATM",{"id":4849,"room":4850,"start":3505,"end":3732,"language":70,"track":4851,"speakers":4853,"zh":4857,"en":4860,"tags":4861,"uri":4862},"JA8LJD",{"en":1154,"zh-hant":10},{"id":286,"name":4852},{"en":288,"zh-hant":288},[4854],{"id":930,"avatar":931,"zh":4855,"en":4856},{"name":933,"bio":934},{"name":933,"bio":934},{"title":4858,"describe":4859,"type":30},"Making Postgres Central in Your Data Center","Postgres has the unique ability to act as a powerful data aggregator in many data centers. This talk shows how Postgres's extensibility, access to foreign data sources, and ability to handle NoSQL-like and data warehousing workloads gives it unmatched capabilities to function in this role.",{"title":4858,"describe":4859,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FJA8LJD",{"id":4864,"room":4865,"start":4866,"end":4867,"language":96,"track":4868,"speakers":4870,"zh":4878,"en":4881,"tags":4884,"uri":4885},"JFALUS",{"en":407,"zh-hant":10},"2026-08-09T15:15:00+08:00","2026-08-09T15:45:00+08:00",{"id":410,"name":4869},{"en":412,"zh-hant":412},[4871],{"id":4872,"avatar":4873,"zh":4874,"en":4877},"DD8WRB","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FDD8WRB_sje3Tun.webp",{"name":4875,"bio":4876},"Leo Kuo","設計背景出身的前端工程師，主要使用 Angular、React 與現代 Web 技術開發產品。平常關注前端架構、開發者體驗、互動介面與效能優化，近年持續探索 Web 技術在遊戲開發上的應用。\r\n目前為獨立遊戲開發者，持續開發《終局棒球》。過去也曾參與 js13kGames、Gamedev.js Jam 等 Web 遊戲開發挑戰。本次分享整理遊戲實際開發過程中的架構判斷、技術取捨，以及遇到的問題與解法。\r\n\r\nA design-background frontend engineer turned indie game developer. Previously built products with Angular and React, with a focus on frontend architecture, developer experience, and performance. Now working full-time on Final Inning Baseball, an indie baseball game.",{"name":4875,"bio":4876},{"title":4879,"describe":4880,"type":30},"不只是 Canvas：React + Phaser 的前端遊戲架構實戰","前端技術不只能做網站，也能成為製作 2D 遊戲的完整工具鏈。但如果 Unity \u002F Godot 已經能跨平台，Phaser 也能直接做遊戲，為什麼還要把 React、Redux、Capacitor 放進遊戲架構裡？\r\n\r\n本議程將分享如何使用 React、Phaser 與現代前端工具，打造一款可在 Web 與 Mobile 上運行的遊戲產品。Phaser 擅長處理遊戲核心，例如 scene、sprite、animation、input 與 game loop；React 則更適合處理遊戲外層的產品介面，例如選單、設定、狀態顯示、modal、流程切換與 responsive UI。\r\n\r\n當遊戲不只是 canvas 裡的一段互動，而是包含多個畫面、流程、資料與產品體驗時，如何切分 Phaser 與 React 的責任邊界，就會成為架構設計的關鍵。本分享會討論哪些狀態應該留在 Phaser，哪些狀態適合交給 React \u002F Redux 管理，這套架構如何延伸到 Capacitor 包裝的手機 App，以及 Web stack 相對 Unity \u002F Godot 在開發速度、測試、自動化、部署與產品整合上的取捨。",{"title":4882,"describe":4883,"type":30},"Beyond Canvas: Frontend Game Architecture with React and Phaser","Frontend technologies are no longer limited to building websites. They can also form a practical toolchain for developing 2D games. This talk uses Final Inning Baseball as a case study to explore how React, Phaser, Redux, and modern frontend tools can be used to build a game product that runs on both Web and Mobile.\r\n\r\nPhaser is responsible for the core game layer, including scenes, sprites, animation, input, and the game loop. React, on the other hand, is better suited for the application layer around the game, such as menus, settings, modals, UI states, flow control, and responsive interfaces.\r\n\r\nWhen a game becomes more than an interaction inside a canvas and starts to include multiple screens, data flows, persistence, and product-level user experience, the boundary between Phaser and React becomes an important architectural decision. This talk discusses what state should stay inside Phaser, what state should be managed by React \u002F Redux, how the architecture can be extended to mobile apps with Capacitor, and how this Web-based stack compares with Unity \u002F Godot in terms of development speed, testing, automation, deployment, and product integration.",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FJFALUS",{"id":4887,"room":4888,"start":3549,"end":3776,"language":328,"track":4889,"speakers":4891,"zh":4899,"en":4902,"tags":4903,"uri":4904},"JFDTVX",{"en":922,"zh-hant":10},{"id":330,"name":4890},{"en":332,"zh-hant":332},[4892],{"id":4893,"avatar":4894,"zh":4895,"en":4898},"JSPN9Q","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FJSPN9Q_ik7RBul.webp",{"name":4896,"bio":4897},"Aaditya Singh","I am Aaditya Singh, the community lead of Ubuntu Nepal and GNOME Nepal. I am also a proud Ubuntu member and a foundation member of the GNOME Foundation.",{"name":4896,"bio":4897},{"title":4900,"describe":4901,"type":30},"Rebuilding Ubuntu Nepal: A Playbook for Reviving Open Source Communities","This session shares the real journey of how the GNOME Nepal community successfully revived Ubuntu Nepal after more than a decade of inactivity. It will walk through the challenges of restarting a dormant open-source community, including rebuilding trust, finding contributors, and creating momentum from zero.\r\n\r\nThe talk will break down the exact strategies used, community outreach, event-driven growth, collaboration between GNOME and Ubuntu ecosystems, and leadership structure. It will also highlight mistakes, lessons learned, and what actually worked in the Nepali context.\r\n\r\nFinally, the session will translate this experience into a practical framework that attendees can apply to revive or bootstrap Ubuntu Circle and similar communities in their own regions.\r\n\r\nMore info: https:\u002F\u002Fevents.canonical.com\u002Fevent\u002F146\u002Fcontributions\u002F943\u002F",{"title":4900,"describe":4901,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FJFDTVX",{"id":4906,"room":4907,"start":3487,"end":4908,"language":70,"track":4909,"speakers":4911,"zh":4915,"en":4918,"tags":4921,"uri":4922},"JLQPTY",{"en":1253,"zh-hant":10},"2026-08-09T10:15:00+08:00",{"id":901,"name":4910},{"en":903,"zh-hant":904},[4912],{"id":2511,"avatar":2512,"zh":4913,"en":4914},{"name":2514,"bio":2515},{"name":2514,"bio":2515},{"title":4916,"describe":4917,"type":30},"When AI Helps, and Why I Still Write the PR Myself","AI tools are making it easier than ever to generate code, explore ideas, and open pull requests quickly. But faster output does not always lead to better PRs. In practice, AI-generated changes often come with familiar problems: larger diffs, weaker context, less clear intent, and higher review overhead.\r\n\r\nIn this talk, I will share a practical view of where AI genuinely helps in day-to-day development, where it starts to create problems, and why I still choose to manually craft most pull requests. Rather than arguing against AI, this session focuses on a more useful question for engineers and open-source contributors: how do we use AI as a tool without giving up ownership, clarity, and reviewer empathy?\r\n\r\nI will also briefly use my own open-source contribution experience, including work related to Zed, as an example to show why a reviewable, well-structured change is often more valuable than a quickly generated one. \r\n\r\nThe goal of this talk is to provide practical suggestions about how to build a proper PR to contribute to open-source in a healthy way, and understand why keep context understanding personally is important.",{"title":4919,"describe":4920,"type":30},"AI 很方便，但有些 PR 我還是想自己寫","這場分享會討論 AI 在軟體開發中何時真正有幫助、何時會帶來審查上的問題，以及為什麼我仍然選擇大多數 PR 由自己撰寫。",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FJLQPTY",{"id":4924,"room":4925,"start":3567,"end":4700,"language":96,"track":4926,"speakers":4928,"zh":4932,"en":4935,"tags":4936,"uri":4937},"JLX3WJ",{"en":407,"zh-hant":10},{"id":410,"name":4927},{"en":412,"zh-hant":412},[4929],{"id":3332,"avatar":3333,"zh":4930,"en":4931},{"name":3335,"bio":3336},{"name":3335,"bio":3336},{"title":4933,"describe":4934,"type":59},"From Word & Excel to JavaScript: Transition Learning Programming Concepts","You already use Word and Excel every day. Turns out, those familiar tools share more with programming than you'd expect.This workshop starts exactly where you are — no coding background needed. We'll walk through how spreadsheet thinking connects to JavaScript, how document layout connects to HTML and CSS, and how a simple interactive button opens the door to modern web frameworks.Everything runs live in the browser. You'll see it, click it, and change it yourself.",{"title":4933,"describe":4934,"type":59},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FJLX3WJ",{"id":4939,"room":4940,"start":3754,"end":3590,"language":96,"track":4941,"speakers":4943,"zh":4954,"en":4957,"tags":4958,"uri":4959},"JSMGDZ",{"en":684,"zh-hant":10},{"id":686,"name":4942},{"en":688,"zh-hant":689},[4944,4947],{"id":2156,"avatar":2157,"zh":4945,"en":4946},{"name":2159,"bio":2160},{"name":2159,"bio":2160},{"id":4948,"avatar":4949,"zh":4950,"en":4953},"FW9HYA","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FFW9HYA_615Ds8f.webp",{"name":4951,"bio":4952},"Denken Chen","Over ten years of experience in software development and writing. My personal website: https:\u002F\u002Fdenkeni.org",{"name":4951,"bio":4952},{"title":4955,"describe":4956,"type":30},"年齡驗證、數位監控、隱私權，討論這些以前，不妨先開源吧","年齡驗證與網路自由一定會兩極對立嗎？線上年齡驗證為國際法規發展趨勢，唯其議題複雜性高，倉促強制推行或未評估技術方案可行性，皆可能造成無效的政策或無益於新興密碼學技術發展。本議程將分析關於數位身分驗證的技術方案，尤其是歐盟正開源發展之年齡驗證 App，以利政策制定者、倡議者或反對者，共同釐清年齡驗證與網路自由的邊界。",{"title":4955,"describe":4956,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FJSMGDZ",{"id":4961,"room":4962,"start":4963,"end":4536,"language":96,"track":4964,"speakers":4966,"zh":4974,"en":4977,"tags":4980,"uri":4981},"JTPCAZ",{"en":1253,"zh-hant":10},"2026-08-09T10:20:00+08:00",{"id":901,"name":4965},{"en":903,"zh-hant":904},[4967],{"id":4968,"avatar":4969,"zh":4970,"en":4973},"FNJS9J","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FFNJS9J_s4UQLeS.webp",{"name":4971,"bio":4972},"Yiwei Ho","我是 Yiwei，來自台中的 AI 工程師。\r\n熱愛開發、side project 與技術分享，目前擔任 Raycast 台灣大使。\r\n歡迎一起交流 AI、開發工具與有趣的產品想法！\r\n我的個人網站：https:\u002F\u002F1wei.dev",{"name":4971,"bio":4972},{"title":4975,"describe":4976,"type":30},"open-slide：從騎車時的靈感到衝上 GitHub Trending","open-slide 是一個專為 AI Agent 設計的開源簡報框架，讓 coding agent 用 React 寫簡報，而非被模板限制。\r\n\r\n**為什麼開發 open-slide？**\r\n\r\n現有 AI 簡報工具（Gamma、Canva、NotebookLM 等）多走「模板＋填字」或直接生成圖像路線，AI 生成文字塞進固定版型，好看但千篇一律，設計自由度低。open-slide 換了一個思路：簡報本質是視覺化的程式碼，Agent 擅長寫程式碼，所以 open-slide 提供的是 slide 的 runtime，讓 Agent 在簡報限制的範圍內，完全自由地生成視覺與資訊內容。\r\n\r\n**這個專案教我的四件事**\r\n\r\n1. 記下每一個靈感：隨手記錄的念頭，就是未來作品的種子\r\n2. 先做再想：動手本身就是最好的解法\r\n3. 主動讓作品被看見：Threads 經營本地能見度，X 接觸國際社群\r\n4. 勇敢用英文交流：一則開源貢獻可能成為最重要的連結\r\n\r\n如果你也有一個放在心裡很久的靈感，希望這場分享能成為你按下「新增專案」的推力！",{"title":4978,"describe":4979,"type":30},"open-slide: From Idea to GitHub Trending","open-slide is an open-source slide framework designed for AI agents, letting coding agents build slides as React components instead of being constrained by templates.\r\n\r\n**Why open-slide?**\r\n\r\nMost AI presentation tools (Gamma, Canva, NotebookLM, etc.) follow a \"template + fill-in\" or direct image generation approach — AI generates text and drops it into fixed layouts. The results look polished but generic, with little room for creative freedom. open-slide takes a different approach: slides are essentially visual code, and agents are great at writing code. So instead of templates, open-slide provides a slide runtime, giving agents full freedom to generate visual and informational content within the constraints of a presentation.\r\n\r\n**Four Things This Project Taught Me**\r\n\r\n1. Capture every idea: A quick note today can become your next project\r\n2. Build first, think later: Taking action is often the best way to find answers\r\n3. Proactively make your work visible: Threads for local reach, X for the global dev community\r\n4. Be brave enough to communicate in English: One open-source contribution might become your most important connection\r\n\r\nIf you've been sitting on an idea for a while, I hope this talk gives you the push to finally hit \"New Project.\"",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FJTPCAZ",{"id":4983,"room":4984,"start":3567,"end":3568,"language":70,"track":4985,"speakers":4987,"zh":4995,"en":4998,"tags":4999,"uri":5000},"KCTQGA",{"en":93,"zh-hant":10},{"id":3467,"name":4986},{"en":3469,"zh-hant":3469},[4988],{"id":4989,"avatar":4990,"zh":4991,"en":4994},"YUKNFG","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002Fagl-prof_oEWdF3X.jpg",{"name":4992,"bio":4993},"Naoto Yamaguchi","Doctor of Informatics.  Developer of embedded Linux for automotive since 2011. AGL and AGL Japan local community member. \r\nHe is a top level contributor for Automotive Grade Linux.  He developed to Linux container based integrated system that is building fully open-source software.  That developed since 2019.\r\nHe is a member of AGL SDV Expert Group now.  He collaborate to many member to realize open-source accelerated SDV(Software Defined Vehicle).",{"name":4992,"bio":4993},{"title":4996,"describe":4997,"type":30},"Mission impossible? How to realize a reproducible and a flexible system integration method for the SoDeV","The SoDeV is an open-source reference Software Defined Vehicle (SDV)  platform.  It's developed by the Automotive Grade Linux (AGL) SoDev project.  It aims to de-fact standard for the SDV platform.\r\nThe SoDeV platform is built on a hypervisor-based host environment and hardware decoupled guest VMs with Virt IO infrastructures.  The example of those guest VMs are AOSP, AGL In Vehicle Infotainment(IVI), Instrument Cluster (IC), gateway, real-time Linux-based runtime environment and RTOS-based runtime environment.\r\nFrom our point of view, \"reproducible\" is a high-priority topic.  From the SoDeV developer's point of view, we need a method for the same environment creation to share the common development environment.  From the SoDeV user point of view, we need to reproduce the SoDeV platform in our own environment to evaluate, modify and more.\r\nThe AGL has a common reproducible integration tool called Yocto.  It already has a host and guest co-build method called multiconfig.  On the other hand, that requires complex configuration.  It's a big barrier for the SoDeV developer and user.\r\nIn this session, I'm talking about an idea for how to fix this issue.  This talk includes my investigation, trial and consideration.  It's common knowledge outside the automotive field.  Because the same issue exists in IoT, Industrial and other fields.",{"title":4996,"describe":4997,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FKCTQGA",{"id":5002,"room":5003,"start":3732,"end":3991,"language":328,"track":5004,"speakers":5006,"zh":5014,"en":5017,"tags":5018,"uri":5019},"KKW9LR",{"en":899,"zh-hant":10},{"id":330,"name":5005},{"en":332,"zh-hant":332},[5007],{"id":5008,"avatar":5009,"zh":5010,"en":5013},"7V3QRQ","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002F7V3QRQ_rAIAaiH.svg",{"name":5011,"bio":5012},"Vutukuri Sreenivas","V Sreenivas is a DevRel, specializing in cybersecurity, Linux security, and cloud-native security. With a strong background in security engineering and open-source contributions, he has worked extensively on security observability and enforcement mechanisms. Passionate about sharing knowledge, he has spoken at various security conferences and actively contributes to the security community. https:\u002F\u002Fwww.linkedin.com\u002Fin\u002Fv-sreenivas-985088203\u002F",{"name":5011,"bio":5012},{"title":5015,"describe":5016,"type":30},"Zero-Trust Workloads with Confidential Containers on Ubuntu","In terms of protecting sensitive data during execution across diverse cloud environments, typical **container isolation** mechanisms frequently fall short. While traditional security secures data at rest and in transit, the **\"data-in-use\"** gap remains leaving memory and running processes vulnerable to a compromised host, hypervisor, or rogue administrator. Without requiring changes to application code, Confidential Containers (CoCo) is a potent open-source technology that makes it possible to isolate workloads safely and effectively using hardware-based Trusted Execution Environments (TEEs). This session examines how memory encryption, remote attestation, and hardware-level isolation offered by CoCo might improve zero-trust security on Ubuntu systems. Measures like standard Linux namespaces and cgroups are less efficient against contemporary infrastructure-level attacks as a result of the sophistication of modern cloud-native threats. By offering a cryptographic method of isolating system memory without compromising cloud flexibility, CoCo tackles these issues. And CoCo is a crucial tool for contemporary data privacy and compliance since it can protect proprietary code, AI models, and sensitive payloads by executing them inside hardware-attested utility VMs on the Linux kernel.\r\n\r\nThe talk will focus on the integration of **Confidential Containers with Ubuntu**, highlighting Canonical’s and the broader open-source community's ecosystem support for TEE-backed virtualization such as Intel TDX and AMD SEV-SNP. Will also include **practical demonstrations** of CoCo deployments on Ubuntu, showcasing how to use the open-source CoCo Operator for running encrypted pods, managing remote attestation, and preventing unauthorized host-level access to container memory. We will walk through real-world examples of deploying hardware-isolated workloads using CNCF tools such as Kata Containers and the Confidential Containers project alongside Ubuntu's MicroK8s, demonstrating how they enforce a true zero-trust architecture in modern cloud infrastructures.\r\n\r\nParticipants will have a thorough grasp of how to use Confidential Containers to improve data-in-use security in Ubuntu environments by the end of this session. They will acquire hands-on experience in deploying CoCo for memory encryption, hardware-based workload isolation, and cryptographic attestation, allowing them to leverage this potent technology in their own infrastructure strategies and actively champion advanced cloud security within the Ubuntu ecosystem.",{"title":5015,"describe":5016,"type":30},[402],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FKKW9LR",{"id":5021,"room":5022,"start":4830,"end":4831,"language":96,"track":5023,"speakers":5025,"zh":5033,"en":5036,"tags":5039,"uri":5040},"KM8X9S",{"en":9,"zh-hant":10},{"id":3645,"name":5024},{"en":3647,"zh-hant":3648},[5026],{"id":5027,"avatar":5028,"zh":5029,"en":5032},"TPJARR","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FTPJARR_uYsrRcN.png",{"name":5030,"bio":5031},"吳承翰 (吳他)","- 喜愛系統程式\r\n- 喜歡教學與寫作",{"name":5030,"bio":5031},{"title":5034,"describe":5035,"type":30},"開始貢獻 Linux Kernel 的第一步，關於申請 Linux Kernel Mentorship Program 經驗談","The Linux Foundation: The Linux Kernel Mentorship Program (以下簡稱 LKMP) 為 The Linux Foundation 主辦的實習計畫，目標是引導對 Linux Kernel 有興趣的開發者開始參與 Linux Kernel 開發與社群貢獻。\r\n\r\n在本場議程中將會詳細介紹 LKMP 實習制度，如何進行申請，歷年來申請的錄取率等等資訊，最後簡單介紹我在 LKMP 計畫中所獲得到的幫助。\r\n\r\n希望對於 LKMP 實習或是對 Linux Kernel 有興趣的聽眾在聽完本場議程之後，能夠獲得關於 LKMP 的詳細資訊，以及了解如何進行準備。最後成功參與 LKMP，逐步加入到 Linux Kernel 社群中。",{"title":5037,"describe":5038,"type":30},"Taking the First Step Into the Linux Kernel World: My Experience Applying to the Linux Kernel Mentorship Program","The Linux Foundation’s Linux Kernel Mentorship Program, hereafter referred to as LKMP, is an internship program organized by The Linux Foundation. The program is designed to help developers who are interested in the Linux kernel take their first steps into kernel development and open-source community contribution.\r\n\r\nIn this talk, I will introduce the LKMP program in detail, including how the internship is structured, how to apply, what applicants need to prepare, and what the historical acceptance rates look like. I will also briefly share my own experience in the program and the support I received through LKMP.\r\n\r\nThis session is intended for people who are interested in applying to LKMP or getting started with Linux kernel development. By the end of the talk, attendees should have a clearer understanding of what LKMP is, how to prepare for it, and how the program can serve as a practical entry point into the Linux kernel community.",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FKM8X9S",{"id":5042,"room":5043,"start":3622,"end":5044,"language":96,"track":5045,"speakers":5047,"zh":5051,"en":5054,"tags":5057,"uri":5058},"KSWEDB",{"en":263,"zh-hant":10},"2026-08-09T14:25:00+08:00",{"id":44,"name":5046},{"en":46,"zh-hant":47},[5048],{"id":4264,"avatar":4265,"zh":5049,"en":5050},{"name":4267,"bio":4268},{"name":4267,"bio":4268},{"title":5052,"describe":5053,"type":59},"瀏覽器追蹤技術、反追蹤策略和使用者自主","現代網頁瀏覽器提供許多看似不屬於網頁該有的 API （功能）：WebGL 讓網頁可以使用圖形加速程式語言、WebAudio 讓網頁可以直接執行複雜的音訊合成指令、WebAssembly 則是允許網頁直接執行低階指令碼。\r\n\r\n這些 API 讓瀏覽器可以做的事情變多了，以往僅能由原生應用程式提供的功能在當今許多都可以由網頁提供。不過這些與硬體緊密結合的 API 讓網頁得以收集更多使用者資料，進而更精準地**跟蹤和識別使用者**。\r\n\r\n不幸的是，**追蹤使用者**正是大多網頁使用這些 API 的方式，而非使用這些 API 提供的獨特硬體整合功能。這些 API 被整合進瀏覽器裡面，網頁要求時自動啟用——從來沒問過使用者的意見。\r\n\r\n本議程內我將分享我對瀏覽器指紋 （browser fingerprint)、反追蹤的簡單研究，會提到 Tor Browser, Brave, VPN 等軟體和技術。並探討反追蹤的動機和策略，介紹一些潛在解決方案，並分享我的使用心得。",{"title":5055,"describe":5056,"type":59},"Browser fingerprinting, anti-fingerprinting and user autonomy","Modern browsers offer a lot of APIs (features) that are tightly integrated with hardware: WebGL, WebAudio, WebAssembly, to name a few.\r\n\r\nHowever, most webpages that use these APIs do not use them for their intended purpose (such as to show 3D graphics, manipulate audio signals, run near-native bytecode for accelerated processing.) Instead, most webpages use these APIs to track, identify and surveil users.\r\n\r\nIn this talk, I will share my research on browser fingerprinting technologies, anti-fingerprinting techniques and strategies employed by major browsers. Finally, I will share potential solutions that I have found and built.",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FKSWEDB",{"id":5060,"room":5061,"start":3567,"end":3568,"language":96,"track":5062,"speakers":5064,"zh":5072,"en":5075,"tags":5078,"uri":5079},"KXANFQ",{"en":3731,"zh-hant":10},{"id":3734,"name":5063},{"en":3736,"zh-hant":3736},[5065],{"id":5066,"avatar":5067,"zh":5068,"en":5071},"M7NR8P","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FM7NR8P_cXLvLbZ.webp",{"name":5069,"bio":5070},"黃國豪","我是 Dennis ，一名後端工程師，目前專注在 Datafusion 及分散式系統，程式之餘喜歡打排球。\r\n希望能在開源的世界留下足跡。",{"name":5069,"bio":5070},{"title":5073,"describe":5074,"type":30},"用 Go 讀懂 Disruptor：從源碼理解高效能 Ring Buffer 的設計哲學","當你的 Go 服務開始出現 latency Spike，第一個懷疑對象往往是 channel。\r\nChannel 內部的 mutex、GC 壓力、false sharing，這些問題在高吞吐場景下會被放大。\r\nLMAX Disruptor 是金融交易系統中解決這類問題的經典方案，但它真的適合 Go 嗎？\r\n本議題帶你從源碼出發，一起讀懂 smarty\u002Fgo-disruptor 的設計，理解 Disruptor 為什麼快，以及在 Go 的生態下它的邊界在哪裡。\r\n\r\n1. Channel 的成本從哪裡來？\r\n用 benchmark 展示 channel 在多核環境下的瓶頸，從 CPU cache line、false sharing 與 mutex contention 建立直覺，讓後續的源碼解析有具體的問題背景。\r\n\r\n2. 讀源碼：Ring Buffer 的核心設計\r\n逐段解析 smarty\u002Fgo-disruptor 的關鍵檔案， sequence.go 的 cache line padding 如何消滅 false sharing、sequencer.go 的 atomic 操作如何取代 mutex、producer\u002Fconsumer gating 機制如何在不加鎖的情況下保證順序。\r\n\r\n3. Go Memory Model 的關鍵轉折\r\nGo 1.19 正式明確了 atomic 操作的 happens-before 語義，這讓 smarty\u002Fgo-disruptor 從 pre-release 進入穩定狀態。我們會從源碼理解這個語義保證在實作上意味著什麼，以及為什麼在這之前 lock-free 的 Go 程式碼其實站在不穩固的地基上。\r\n\r\n4. 邊界在哪裡：什麼時候 Channel 就夠了\r\n以 exchange matching engine 的設計評估為例，說明在真實系統中如何判斷是否需要 Disruptor:吞吐量需求、團隊維護成本、系統複雜度的三角取捨。\r\n\r\n5. 從 Disruptor 借回 Go 的設計概念\r\n即使不用 Disruptor，它的設計思想仍然可以改善你的 Go 系統：預分配 slice 減少 GC 壓力、struct padding 避免 false sharing、ring buffer 作為固定大小的 event queue。\r\n\r\n- 目標受眾： 對 Go 並發有基礎認識、對系統效能調優或底層設計感興趣的工程師。不需要有 Disruptor 或金融系統背景。\r\n- 帶走什麼： 讀源碼的方法、對 Go memory model atomic 語義的具體理解、一個判斷是否需要 Disruptor 的決策框架。\r\n所有參考內容皆為公開開源專案，技術方法不依賴任何特定雲端平台或商業工具。",{"title":5076,"describe":5077,"type":30},"Reading the Disruptor in Go: Understanding High-Performance Ring Buffer Design Through Source Code","When your Go service starts showing latency spikes, channels are often the first suspect. The mutex inside channel, GC pressure, and false sharing . These costs get amplified in high-throughput scenarios. \r\n\r\nThe LMAX Disruptor is a classic solution from financial trading systems, but does it actually fit Go?\r\nThis talk walks through the source code of smarty\u002Fgo-disruptor together, understanding why the Disruptor is fast and where its boundaries lie in the Go ecosystem.\r\n\r\n1. Where do channel costs come from?\r\nBenchmarks show channel bottlenecks in multi-core environments, building intuition around CPU cache lines, false sharing, and mutex contention.\r\n\r\n2. Reading the source: core ring buffer design\r\nWalking through key files: how sequence.go's cache line padding eliminates false sharing, how sequencer.go's atomic operations replace mutexes, and how producer\u002Fconsumer gating guarantees ordering without locks.\r\n\r\n3. The Go Memory Model turning point\r\nGo 1.19 formally clarified happens-before semantics for atomic operations, moving smarty\u002Fgo-disruptor from pre-release to stable. We'll read the source to understand what this guarantee means in practice, and why lock-free Go code before this was standing on shaky ground.\r\n\r\n4. Where's the boundary: when are channels good enough?\r\nUsing a betting exchange matching engine evaluation as a case study: a practical decision framework balancing throughput requirements, team maintenance cost, and system complexity.\r\n\r\n5. Borrowing Disruptor ideas back into Go\r\nEven without using the Disruptor, its design principles can improve your Go systems: pre-allocated slices to reduce GC pressure, struct padding to avoid false sharing, ring buffers as fixed-size event queues.",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FKXANFQ",{"id":5081,"room":5082,"start":3643,"end":3549,"language":96,"track":5083,"speakers":5085,"zh":5093,"en":5096,"tags":5099,"uri":5100},"KZ9PTY",{"en":237,"zh-hant":10},{"id":241,"name":5084},{"en":243,"zh-hant":243},[5086],{"id":5087,"avatar":5088,"zh":5089,"en":5092},"WEYURA","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FICON_512_Q7tZ8Yl.png",{"name":5090,"bio":5091},"蒼時弦也","現為 AI 工程師，今年正在思考要不要練習一些傳統技藝，像是手寫程式之類的？",{"name":5090,"bio":5091},{"title":5094,"describe":5095,"type":30},"讓 AI 接管你的應用，打造微秒級無縫的 Ruby 安全沙盒","在 AI 時代中，過去軟體開發的方式有了很多改變，也帶來非常多不同的契機。你是否想過能在 Ruby 或者 Rails 的環境中，透過 AI 動態地實作新功能？\r\n\r\nKobako 原本是為了在 Ruby 中實現 Cloudflare 提出的 Code Mode 機制，讓 AI 直接撰寫腳本替代大量的 MCP 工具，所設計的 WebAssembly 沙盒（Sandbox）\r\n\r\n然而，受到 druby 的啟發，Kobako 能夠利用 Ruby 的語言特性，動態與宿主（Host）環境的任意物件進行無縫互動，反而讓 AI 能夠以完全不同的形式來協助軟體開發。\r\n\r\n沙盒技術正是今年被大量討論的「駕馭工程（Harness Engineering）」其中一項重要支線。這場演講將圍繞著「如何安全運行程式碼」的核心，分享 Kobako 的開發與探索過程，展現一個善用 Ruby 語言特性而生的安全沙盒。",{"title":5097,"describe":5098,"type":30},"Let AI Take Over Your App: Building a Seamless, Microsecond-Level Ruby Security Sandbox","In the era of AI, traditional software development paradigms have shifted significantly, bringing forth entirely new opportunities. Have you ever wondered if AI could dynamically implement new features right within your Ruby or Rails environment?\r\n\r\nKobako was originally designed as a WebAssembly sandbox to bring the \"Code Mode\" mechanism into the Ruby ecosystem, allowing AI to write short scripts directly instead of orchestrating a massive number of fragmental MCP (Model Context Protocol) tools.\r\n\r\nHowever, inspired by druby, Kobako leverages Ruby’s dynamic language features to interact seamlessly with arbitrary objects in the Host environment. This opens up a completely different approach for AI to assist and drive software development.\r\n\r\nSandbox technology is a crucial branch of \"Harness Engineering,\" a topic being widely discussed this year. Centered on the core challenge of how to execute untrusted code safely, this talk will share the development and exploration journey of Kobako, showcasing a security sandbox born purely from embracing Ruby’s unique language features.",[402],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FKZ9PTY",{"id":5102,"room":5103,"start":3568,"end":3548,"language":70,"track":5104,"speakers":5106,"zh":5114,"en":5117,"tags":5118,"uri":5119},"L8KH3U",{"en":121,"zh-hant":10},{"id":4034,"name":5105},{"en":4036,"zh-hant":4036},[5107],{"id":5108,"avatar":5109,"zh":5110,"en":5113},"EXJJED","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FEXJJED_nBv0xhF.webp",{"name":5111,"bio":5112},"Evan Prodromou","Evan is the Research Director at the Social Web Foundation. He first engaged with the Fediverse when launching the identi.ca project in 2008. He co-authored OStatus and ActivityPub, and chaired the Social Web Working Group at the W3C from 2014 to 2018. Today, he maintains the ActivityPub and Activity Streams specifications, leads efforts to bring end-to-end encryption and private groups to the Fediverse, and builds infrastructure like tags.pub to make the Fediverse better and more interesting. He wrote the book \"ActivityPub: Programming for the Social Web\" (O'Reilly Media, 2024) to help developers understand and use the protocol.",{"name":5111,"bio":5112},{"title":5115,"describe":5116,"type":30},"ActivityPub 1.1","First published by the W3C in January of 2018, ActivityPub has been the protocol that binds the Fediverse together. The hard-won experience of hundreds of developers, who have taken the words of the spec and turned it into a living network, is spread across forums, blog posts, chat rooms and toots. The W3C has authorized a new release of the specification, with updated text, examples, and clarifications. This talk will cover what's coming with ActivityPub 1.1 and what you need to know to get ready for it.",{"title":5115,"describe":5116,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FL8KH3U",{"id":5121,"room":5122,"start":3568,"end":3548,"language":70,"track":5123,"speakers":5125,"zh":5129,"en":5132,"tags":5133,"uri":5134},"LCKXCL",{"en":67,"zh-hant":10},{"id":3993,"name":5124},{"en":3995,"zh-hant":3995},[5126],{"id":1849,"avatar":1850,"zh":5127,"en":5128},{"name":1852,"bio":1853},{"name":1852,"bio":1853},{"title":5130,"describe":5131,"type":30},"Underwater Heritage as an Open Knowledge Graph with Neo4j","## Description\r\n\r\nUnderwater cultural heritage is fragile and often invisible — not only beneath the sea, but buried in PDFs, scattered archives, and disconnected databases. While LLMs can generate summaries, they do not preserve structured, verifiable knowledge over time. Relational tables store records, but struggle to capture complex relationships such as trade routes, ocean currents, artifacts, and cross-border exchange.\r\n\r\nThis talk proposes modeling underwater heritage as an open knowledge graph — not as a product, but as shared, open infrastructure for cultural memory.\r\n\r\nThese ideas align with open knowledge ecosystems such as Wikidata and OpenStreetMap, where structured, linked data enables global collaboration across communities and domains.\r\n\r\nRather than focusing on specific tools, the session introduces open modeling principles and linked data concepts that make knowledge interoperable, extensible, and community-driven. We will also briefly demonstrate a minimal example using Neo4j as one practical implementation.\r\n\r\nTaiwan, like Japan, is shaped by maritime history. What if we could structure this shared heritage as an open, cross-border knowledge graph — collaboratively maintained and globally connected?\r\n\r\n**At its core, this talk asks:**\r\n\r\n**If knowledge must outlive tools, models, and vendors — where should it live?**\r\n\r\nThis session is for beginners and anyone interested in open data, linked data, and meaningful knowledge design. All examples use open data and open tools, so you can explore and extend the approach yourself.\r\n\r\n## Outline (30 minutes)\r\n\r\n1. Fragmented cultural data  \r\n2. Limits of PDFs, tables, and AI-generated summaries  \r\n3. Open knowledge graph and linked data principles  \r\n4. Minimal example (Neo4j as one implementation)  \r\n5. Cross-border and community-driven knowledge  \r\n\r\n## Key Takeaways\r\n\r\nParticipants will:\r\n\r\n- Understand why current approaches fail to capture complex relationships  \r\n- Learn how to model heritage as interconnected, structured data  \r\n- See a simple, practical graph-based example  \r\n- Understand the role of open, linked knowledge in global and cross-border collaboration",{"title":5130,"describe":5131,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FLCKXCL",{"id":5136,"room":5137,"start":3567,"end":3568,"language":96,"track":5138,"speakers":5140,"zh":5148,"en":5151,"tags":5152,"uri":5153},"LDEJEG",{"en":39,"zh-hant":10},{"id":384,"name":5139},{"en":386,"zh-hant":10},[5141],{"id":5142,"avatar":5143,"zh":5144,"en":5147},"XTDRHG","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FXTDRHG_0x1qbd6.jpg",{"name":5145,"bio":5146},"bing","在區塊鏈社群打雜多年，同時也在金融業做 Web3，主要做 CBDC、支付以及資產代幣化有關之開發與研究。",{"name":5145,"bio":5146},{"title":5149,"describe":5150,"type":30},"Ethereum 101：新手入門與台灣社群指南","1. 區塊鏈不是萬能答案，到底適合解決什麼問題？\r\n2. 區塊鏈這十年，到底證明了什麼？又證偽了什麼？\r\n3. 如果今天重新開始學 Ethereum，新手該怎麼做？又有哪些社群可以加入？\r\n\r\n這場演講不會把 Ethereum 包裝成萬能答案，而是會從新手角度出發，重新整理區塊鏈這些年真正被證明有價值的地方，以及哪些應用只是曾經的想像。前半段將用容易理解的方式說明 Ethereum 的核心概念與限制，幫助聽眾建立正確的基礎知識；後半段則會聚焦在「如何開始學習」，整理台灣在地的 Ethereum 相關社群與學習資源，並結合社群前輩的經驗，分享實際可行的入門路徑與未來值得關注的方向。\r\n\r\n如果你曾經對 Ethereum 好奇，卻不知道它現在還值不值得學、該從哪裡開始、或有哪些人與社群可以一起前進，這場演講希望成為你走進 Ethereum 生態的第一個入口。",{"title":5149,"describe":5150,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FLDEJEG",{"id":5155,"room":5156,"start":3968,"end":3969,"language":70,"track":5157,"speakers":5159,"zh":5163,"en":5166,"tags":5167,"uri":5168},"LDNNEY",{"en":1063,"zh-hant":10},{"id":1065,"name":5158},{"en":1067,"zh-hant":1068},[5160],{"id":1849,"avatar":1850,"zh":5161,"en":5162},{"name":1852,"bio":1853},{"name":1852,"bio":1853},{"title":5164,"describe":5165,"type":30},"From SPY×FAMILY to Evolving Knowledge Graphs: Tracking Wikipedia Changes with Neo4j and LLMs","Many of us know SPY×FAMILY — a “family” where each member has a hidden identity. At first, it looks like a simple family story, but behind it is a network of hidden relationships.\r\n\r\nWhat if we could visualize such hidden relationships in knowledge?\r\n\r\nIn the open-source world, we have access to a huge amount of unstructured data: Wikipedia, documents, and web pages. These sources are easy to read, but difficult to reuse, connect, and track as their content changes over time. Relationships may be hidden in text, but their importance, context, and changes over time are even harder to see.\r\n\r\nThe core of this session is a step-by-step demonstration of how to turn a Wikipedia page into a knowledge graph using Neo4j and the Neo4j LLM Knowledge Graph Builder. The demonstration uses LLMs to extract entities and relationships from the Wikipedia page, and stores the extracted knowledge as a graph that can be explored, compared, and reused.\r\n\r\nThen we will pick two versions of the same Wikipedia page — past and present — and compare them as graphs. This allows us to see how knowledge evolves: what was added, what changed, and how relationships grow over time.\r\n\r\nThis session is for developers, data engineers, and open-source contributors who want to build their own knowledge graph environment. By following the steps, participants will learn how to start from open data, build a graph, compare versions, and apply the same approach to their own documents or web content.\r\n\r\nThis talk is not only about building graphs. It is about exploring knowledge, comparing it, and understanding how it grows.\r\n\r\nEverything in this session is based on open data and open tools. No special dataset is required — just use the data already around you.",{"title":5164,"describe":5165,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FLDNNEY",{"id":5170,"room":5171,"start":3504,"end":3505,"language":96,"track":5172,"speakers":5174,"zh":5182,"en":5185,"tags":5188,"uri":5189},"LJ9TKQ",{"en":67,"zh-hant":10},{"id":3993,"name":5173},{"en":3995,"zh-hant":3995},[5175],{"id":5176,"avatar":5177,"zh":5178,"en":5181},"GDQ3EW","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FGDQ3EW_nUulOJF.jpg",{"name":5179,"bio":5180},"洪哲賢","畢業於國際商務溝通碩士，長期與 Wikidata Taiwan 社群合作的自由譯者，對各類科技都有著高度興趣。目前正在失業中，請多多指教。",{"name":5179,"bio":5180},{"title":5183,"describe":5184,"type":30},"維基數據資料嵌入專案，讓 AI 讀懂維基百科。","我們將會介紹來自 德國維基媒體 為了應對如今 AI 世代所提出的最新的計畫之一：維基數據嵌入專案，這個專案有著相當單純的目標，要讓 AI 讀懂維基。維基數據作為如今資訊社會最重要的基石之一，其重要性來自於其龐大且自由開放的結構化數據，為資訊自由流通做出了莫大的貢獻。如今面對來自 AI 的全新挑戰，維基媒體嵌入專案要做的就是讓維基數據的資料超越機器可讀的標準，使 AI 系統也能自由且正確的使用其中的資訊。",{"title":5186,"describe":5187,"type":30},"Wikidata Embedding Project - Building the Wikipedia for AI agents","In this talk, we will introduce the \"Wikidata Embedding Project\", a project launched 2025 October 1st. This project is one of the most important responses to the era of AI from the Wiki communities. Wikidata Embedding project has a simple goal: to bring the next evolution of Wikidata into life. Wikidata is one of the most important infrastructures in the modern age; it curates a massive collection of structured data, which can be easily accessed by machines, and makes it freely available to anyone. Wikidata Embedding project aims to take it one step further, not only machine-readable, but also making it AI-readable by embedding the structured dataset into a vector dataset.",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FLJ9TKQ",{"id":5191,"room":5192,"start":3708,"end":3709,"language":96,"track":5193,"speakers":5195,"zh":5203,"en":5206,"tags":5207,"uri":5208},"LSRELU",{"en":192,"zh-hant":10},{"id":196,"name":5194},{"en":198,"zh-hant":198},[5196],{"id":5197,"avatar":5198,"zh":5199,"en":5202},"XTRQDX","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002Fme_linkedin_rwldHF1.jpeg",{"name":5200,"bio":5201},"Gavin Guo","Gavin Guo is a Linux kernel engineer at Igalia working on the scheduler subsystem (sched_ext \u002F LAVD) and AI-assisted kernel debugging tooling. Outside the kernel, he builds COCA Dashboard, an open-source MCP-native vocabulary-learning platform combining the Corpus of Contemporary American English with the FSRS-4.5 spaced-repetition algorithm. He has spoken at Linux Plumbers Conference 2025 and Open Source Summit Japan 2025, and previously presented kernel work at LPC and OSPM.",{"name":5200,"bio":5201},{"title":5204,"describe":5205,"type":30},"LAVD: From Gaming to Server Fleets — a BPF Scheduler for Cloud, ML, and Edge","scx_lavd is a sched_ext scheduler implemented in BPF — dynamically loadable at runtime without a kernel rebuild. It started in 2024 as a Linux gaming experiment for Valve's Steam Deck, and has since matured into a serious contender for server workloads. At Linux Plumbers Conference 2025, Meta engineers presented LAVD as a candidate default scheduler for their server fleet. This session would reveal its core design ideas and how recent development benefits cloud and server workloads — such as the `cpu.max` cgroup controller, IRQ-aware task steering, latency criticality propagation, and the task-size-aware load balancer. Beyond cloud, the same design supports power-aware mobile\u002Fedge deployments and hybrid CPU architectures — making LAVD relevant across cloud, automotive, and mobile vendors.",{"title":5204,"describe":5205,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FLSRELU",{"id":5210,"room":5211,"start":3991,"end":3755,"language":328,"track":5213,"speakers":5215,"zh":5232,"en":5235,"tags":5236,"uri":5237},"M3DGHA",{"en":5212,"zh-hant":10},"TR310-2",{"id":330,"name":5214},{"en":332,"zh-hant":332},[5216,5223,5226],{"id":5217,"avatar":5218,"zh":5219,"en":5222},"G3QJYJ","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002F320561656_682253236672537_9175103212742739951_n_7UgspC5.jpg",{"name":5220,"bio":5221},"Youngbin Han","Youngbin is a long time member of Ubuntu Community - mostly involved with leading Ubuntu Circles in Korea. He led Ubuntu Korea and was a member of Ubuntu LoCo Council previously. and also part of UbuCon Asia Committee.",{"name":5220,"bio":5221},{"id":2765,"avatar":2766,"zh":5224,"en":5225},{"name":2768,"bio":2769},{"name":2768,"bio":2769},{"id":5227,"avatar":5228,"zh":5229,"en":5231},"W3ELZZ","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FW3ELZZ_mNKYiNG.svg",{"name":332,"bio":5230},"UbuCon Asia is a community-organized conference connecting Ubuntu community in Asia.\r\nThis annual event connects enthusiast, engineers, creators, researchers, entrepreneurs and contributors across Asia.\r\nMore info: https:\u002F\u002Fwww.ubucon.asia\u002F",{"name":332,"bio":5230},{"title":5233,"describe":5234,"type":59},"Ubuntu Circles BoF","Interested in running your own Local Ubuntu community (which is called Ubuntu Circles)? Or would like to connect with folks who run Ubuntu Circles in their city? Join this BoF to learn what is it and how to start your own and practices from other Ubuntu Circles leaders.\r\n\r\nMore info: https:\u002F\u002Fevents.canonical.com\u002Fevent\u002F146\u002Fcontributions\u002F930\u002F",{"title":5233,"describe":5234,"type":59},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FM3DGHA",{"id":5239,"room":5240,"start":3549,"end":4830,"language":42,"track":5241,"speakers":5243,"zh":5255,"en":5258,"tags":5261,"uri":5262},"M7AW33",{"en":67,"zh-hant":10},{"id":3993,"name":5242},{"en":3995,"zh-hant":3995},[5244,5247,5250],{"id":2710,"avatar":2711,"zh":5245,"en":5246},{"name":2713,"bio":10},{"name":2713,"bio":10},{"id":3998,"avatar":3999,"zh":5248,"en":5249},{"name":4001,"bio":4002},{"name":4001,"bio":4002},{"id":5251,"avatar":3,"zh":5252,"en":5254},"F7AKW7",{"name":5253,"bio":10},"Assanges",{"name":5253,"bio":10},{"title":5256,"describe":5257,"type":59},"開放內容網站的法治教育講座-匯入政策與案例探討","有鑑於大部分人不熟悉OpenStreetMap與Wikidata的匯入規則，特地運用COSCUP時段來處理，\r\n\r\n以下成功案例：\r\n\r\n- 村里匯入與連結Wikidata與戶役政資訊系統代碼\r\n- 山屋\r\n- 臺北好水飲水點\r\n- 河川相關 (宿霧維基百科的問題)\r\n\r\n以下有狀況的案例：\r\n\r\n技術或是授權問題的案例：\r\n- 村里匯入時參考舊資料，漏掉臺南2018年整併的里\r\n- 宿霧維基百科bot產生條目\u002FWikidata項目的重覆或是整併河川問題\r\n- 登山路徑\r\n- ETC門架\r\n- 環境資訊協會的飲水地圖\r\n\r\n爭議處理機制：\r\n\r\n- 社群共識與執行回退\r\n- OpenStreeetMap Foundation Data Working Group，近期 AED 爭議",{"title":5259,"describe":5260,"type":59},"License Lecture of Open Data Website: Case Study and Import Policy","It seems most people are not familiar with OpenStreetMap and Wikidata import policy, we will host a panel to discuss varies issue during COSCUP\r\n\r\nSuccess Import\r\n\r\n- Village import and linked with WIkidata and Household Register System Code\r\n- Mountain Cabin\r\n- Drinking Water from Taipei City\r\n\r\nCases with Issue:\r\n\r\n- OCF assist 2026 AED Case\r\n- Emergency Shelter\r\n- Psychological clinic\r\n_ Kindergarten\r\n- Parks in Taichung\r\n\r\nTechnical or License Issue\r\n- Hiking Trail\r\n- ETC Toll\r\n- Drinking Water Map from TEIA",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FM7AW33",{"id":5264,"room":5265,"start":3969,"end":4447,"language":96,"track":5266,"speakers":5268,"zh":5276,"en":5279,"tags":5282,"uri":5283},"M8WTHA",{"en":9,"zh-hant":10},{"id":3645,"name":5267},{"en":3647,"zh-hant":3648},[5269],{"id":5270,"avatar":5271,"zh":5272,"en":5275},"R93ZZL","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002F1701582660980_qR5bQ1z.jpeg",{"name":5273,"bio":5274},"voidful","大家好，我是 Eric Lam，一名專注於人工智慧研究與模型開發的博士生，也熱衷在 GitHub 上做開源貢獻。",{"name":5273,"bio":5274},{"title":5277,"describe":5278,"type":30},"透過ClaudeCode，Gemini Cli，Codex來瞭解harness","當我們使用 Claude Code、Gemini CLI 或 OpenAI Codex 這類終端 AI 編碼代理時，表面上只是「打幾個字，模型就幫你改 code」。但背後真正讓這一切運作的，是一個被稱為 Harness 的執行框架——它負責組裝 prompt、調度工具呼叫、管控檔案權限、維護對話歷史，並在人類與 LLM 之間建立信任邊界。\r\n\r\n本演講以實際動手拆解三大主流 coding agent 為主軸。首先，我們從 Gemini CLI 的 REPL 互動、SSE 串流與 MCP 整合出發，理解 harness 如何管理「模型說話 → 工具執行 → 結果回傳」的 Agent Loop；接著透過 OpenAI Codex 的 sandbox 分級與 permission 模型，看 harness 怎麼在「全自動」與「人工確認」之間劃線；最後藉由 Claude Code 的 trust 邊界、session 分支與 skill 系統，探討 harness 如何讓 agent 在跨 session 保持記憶、依指令自我約束。\r\n\r\n為了從閱讀進化到實作，講者將這三套系統的核心設計提煉成一個 Rust 原生專案 GemiClawDex——以不到 5,000 行程式碼，重現 Agent Loop、8 個內建工具、多供應商切換、四級 sandbox、三級 permission prompt、coordinator 多代理協調，以及 MEMORY.md 持久化記憶等關鍵機制。透過這個「從零重寫」的過程，聽眾將能具體理解 harness 的五個核心層：指令載入、prompt 組裝、工具註冊、信任決策、session 持久化，並帶走一份可直接 cargo build 的參考實作。\r\n\r\n無論你是想理解 AI coding agent 為什麼能改你的 code 又不會搞爆你的 repo，還是想自己打造一個，這場演講都能給你一張清晰的架構地圖。",{"title":5280,"describe":5281,"type":30},"Understanding the Harness Through Claude Code, Gemini CLI, and Codex","When we use terminal AI coding agents like Claude Code, Gemini CLI, or OpenAI Codex, it feels like magic — type a few words and the model rewrites your code. But what actually makes this work is an execution framework known as the Harness — responsible for assembling prompts, orchestrating tool calls, enforcing file permissions, maintaining conversation history, and establishing trust boundaries between humans and LLMs.\r\n\r\nThis talk takes a hands-on approach to dissecting three major coding agents. We start with Gemini CLI's REPL interaction, SSE streaming, and MCP integration to understand how the harness manages the \"model speaks → tool executes → result returns\" Agent Loop. Next, through OpenAI Codex's sandbox tiers and permission model, we examine how the harness draws the line between \"full-auto\" and \"human-in-the-loop.\" Finally, via Claude Code's trust boundaries, session branching, and skill system, we explore how the harness enables agents to retain memory across sessions and self-constrain based on instructions.\r\n\r\nTo move from reading to building, the speaker distilled the core designs of all three systems into a Rust-native project called GemiClawDex — reproducing the Agent Loop, 8 built-in tools, multi-provider switching, four-tier sandbox, three-level permission prompts, coordinator-based multi-agent orchestration, and MEMORY.md persistent memory in under 5,000 lines of code. Through this \"rewrite from scratch\" journey, the audience will gain a concrete understanding of the five core layers of a harness: instruction loading, prompt assembly, tool registration, trust decisions, and session persistence — and walk away with a reference implementation they can cargo build on the spot.\r\n\r\nWhether you want to understand why an AI coding agent can modify your code without blowing up your repo, or you want to build one yourself, this talk will give you a clear architectural map.",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FM8WTHA",{"id":5285,"room":5286,"start":5287,"end":3776,"language":96,"track":5288,"speakers":5290,"zh":5298,"en":5301,"tags":5302,"uri":5303},"M9AKQY",{"en":326,"zh-hant":10},"2026-08-09T14:55:00+08:00",{"id":3268,"name":5289},{"en":3270,"zh-hant":3271},[5291],{"id":5292,"avatar":5293,"zh":5294,"en":5297},"QMJRYB","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FQMJRYB_Y4ngkKF.webp",{"name":5295,"bio":5296},"Justin Lee （阿修）","UX 和 iOS 背景的開發者，致力於透過對人的觀察和技術實踐提供最佳的使用體驗。近兩年的重心，在研究如何透過 AI 提升軟體團隊開發效率。",{"name":5295,"bio":5296},{"title":5299,"describe":5300,"type":30},"自己的 AI agent dashboard 自己刻——ccxray 實作筆記","你每天用 Claude Code 寫程式，但真的知道每輪 turn 燒了多少 token、context window 還剩幾趴、cache 有命中嗎？辛苦寫的 skill 到底被呼叫了多少次？\r\nccxray 是一支透明 HTTP proxy + 即時 web dashboard，夾在 Agent 與 API 之間。把所有的 request 和 response 都 log 下來，並提供清楚的視覺化。\r\n本場演講將揭露本專案的開發過程。\r\nhttps:\u002F\u002Fgithub.com\u002Flis186\u002Fccxray",{"title":5299,"describe":5300,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FM9AKQY",{"id":5305,"room":5306,"start":3505,"end":3732,"language":70,"track":5307,"speakers":5309,"zh":5317,"en":5320,"tags":5321,"uri":5322},"MGKSRE",{"en":67,"zh-hant":10},{"id":3993,"name":5308},{"en":3995,"zh-hant":3995},[5310],{"id":5311,"avatar":5312,"zh":5313,"en":5316},"VPLF3Y","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FVPLF3Y_mO2da9U.webp",{"name":5314,"bio":5315},"Alan Ang","I am the Senior Partner Manager for Wikidata at Wikimedia Deutschland, and I've been working on data partnership activities with open-source AI projects, public and government agencies, tech companies, and cultural heritage institutions for close to five years.",{"name":5314,"bio":5315},{"title":5318,"describe":5319,"type":30},"Wikidata MCP: accessing the world's largest free and open knowledge graph via your LLM!","The Wikidata Model Context Protocol (MCP) was launched in 2025. It enables LLM users to access Wikidata's data via natural language prompts. This session introduces the Wikidata Embedding Project, Wikidata MCP, and a brief showcase of the Wikidata MCP in action. Join me to learn how to access the world's largest free and open knowledge graph via LLM.",{"title":5318,"describe":5319,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FMGKSRE",{"id":5324,"room":5325,"start":4140,"end":3643,"language":96,"track":5326,"speakers":5328,"zh":5336,"en":5339,"tags":5340,"uri":5341},"MQN7EY",{"en":167,"zh-hant":10},{"id":3526,"name":5327},{"en":3528,"zh-hant":3528},[5329],{"id":5330,"avatar":5331,"zh":5332,"en":5335},"YEFBBZ","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FYEFBBZ_Of7HP8x.webp",{"name":5333,"bio":5334},"freddy4212","熱愛互動技術應用，總思考著如何運用技術來改善生活、引發議題思考，擅長 TinyML 與 OpenCV 應用......最近迷上 LLM 這個新世界，每天都有追不完的新知識 XDDD",{"name":5333,"bio":5334},{"title":5337,"describe":5338,"type":30},"解救被封印的 PDF—一起來打造臺灣的 FinePDFs 資料集吧","#### 為什麼要開這個議題？\r\n- **資料，乃訓練之基礎—** 為了讓模型更懂臺灣，我們需要你一起貢獻更多更多的臺灣知識！\r\n- **PDF 資料比你想像的有價值—** 有許多重要知識都被封印在 PDF 文件或是圖片之中，這樣的資料無法作為模型訓練內容。\r\n- **告別「工人智慧」輕輕鬆鬆完成貢獻—** 我們設計了一套半自動化的 PDF 文件處理系統，將在本議程介紹如何快速貢獻並取得 Credit！\r\n- **血淚史大公開—** 分享 PDF 處理中的辛酸血淚，讓你快速就能掌握各種撇步！\r\n\r\n#### 誰適合來聽這個議程？\r\n- **知識囤積狂與文獻愛好者—** 手邊一堆珍貴的 PDF 檔案，卻不知如何讓它們發揮最大價值的你。\r\n- **文字潔癖與細節控—** 看到錯字跟排版亂掉就渾身不對勁？我們超需要你來幫忙把關最後一哩路（人工審閱 XDDDD）！\r\n- **開源狂熱份子—** 想了解如何用自動化工具與OCR Model 對抗排版地獄的你。\r\n\r\n#### 這個議題期望帶來什麼樣的影響力？\r\n- **我們是 TwinkleAI—** 一個致力於建立「臺灣在地資料集與模型」的社群，目前已公開多筆資料集以及數顆繁體中文模型。\r\n- **擴張開源宇宙的版圖—** 繼目前進行中的「Twinkle Voice 語音資料集」、「Twinkle Vision 視覺資料集」專案後，我們將在本次議程正式啟動「Twinkle FinePDFs-zhtw」！\r\n- **人人都能成為開源貢獻者—** 透過我們準備好的工具，不管是工程師還是麻瓜，都能輕鬆參與這個大型開源運動。來聽這場議程，一起在臺灣 AI 發展的歷史上留下你的 GitHub ID 與名字吧！",{"title":5337,"describe":5338,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FMQN7EY",{"id":5343,"room":5344,"start":4700,"end":3968,"language":328,"track":5345,"speakers":5347,"zh":5355,"en":5358,"tags":5359,"uri":5360},"MQVP3C",{"en":899,"zh-hant":10},{"id":330,"name":5346},{"en":332,"zh-hant":332},[5348],{"id":5349,"avatar":5350,"zh":5351,"en":5354},"RWWKX3","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FRWWKX3_hyEWrYx.webp",{"name":5352,"bio":5353},"Mitsuya Shibata","Mitsuya Shibata is a member of the Ubuntu Japanese Team and works as an embedded systems engineer and engineering manager at SOUM Corporation. He focuses on low-level system development and practical applications of open-source technologies on Linux.",{"name":5352,"bio":5353},{"title":5356,"describe":5357,"type":30},"Local Real-Time Subtitles on Ubuntu with Moonshine and Pympress","Language barriers remain a significant challenge in international conferences. Not all speakers are fluent in English, and even when they are, their speech may not always be easy for diverse audiences to understand. While some events provide professional captioning or simultaneous interpretation, such services are not always available for every session.\r\n\r\nIn this talk, I will present a self-contained approach that enables speakers to generate real-time subtitles directly from their own laptop. By combining Moonshine Voice, a lightweight and efficient speech-to-text (STT) system that can run even on devices like Raspberry Pi, with Pympress, a PDF-based presentation tool, we can build a fully local, automatic subtitle system on Ubuntu.\r\n\r\nThis approach allows speakers to make their talks more accessible without relying on external infrastructure or services, making it easier to participate in international events regardless of location or available resources.\r\n\r\nMore info: https:\u002F\u002Fevents.canonical.com\u002Fevent\u002F146\u002Fcontributions\u002F935\u002F",{"title":5356,"describe":5357,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FMQVP3C",{"id":5362,"room":5363,"start":3754,"end":3755,"language":70,"track":5364,"speakers":5366,"zh":5374,"en":5377,"tags":5378,"uri":5379},"MUHKX7",{"en":121,"zh-hant":10},{"id":4034,"name":5365},{"en":4036,"zh-hant":4036},[5367],{"id":5368,"avatar":5369,"zh":5370,"en":5373},"7FUJ7A","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002F7FUJ7A_1phPQnE.webp",{"name":5371,"bio":5372},"Jiwon Kwon","Jiwon is a software engineer based in Seoul, interested in open source, the Fediverse, developer tools, and creative technology. She contributes to Fedify and has been exploring new ways to build with ActivityPub, including lightweight and portable Fediverse software. She also sometimes teaches media art and makes projects with computing devices, connecting software, hardware, and artistic practice.",{"name":5371,"bio":5372},{"title":5375,"describe":5376,"type":30},"Feder: One ActivityPub Core, Many Runtimes","Feder is a Rust framework for building ActivityPub applications from a portable protocol core and platform-specific runtimes. The project grew out of my work in the Fedify ecosystem and from a question that kept coming back: what would it take for a Fediverse server to become smaller, cheaper, and more portable than the usual VPS-based web application?\r\n\r\nOne motivating image was a single-user ActivityPub server as a physical device — something you could plug into power and Ethernet. Pushing that idea further led to a more difficult question: could parts of a Fediverse server run on embedded hardware such as ESP32-class boards, closer to sensor-network devices than conventional servers? Feder does not assume that a full Mastodon-like server can simply be moved onto a microcontroller. Instead, it asks how ActivityPub software should be decomposed when different machines have very different resources.\r\n\r\nFeder separates ActivityPub protocol logic from platform execution. The core is responsible for federation behavior such as inbox\u002Foutbox state and delivery decisions, while runtimes take responsibility for networking, storage, clocks, and execution. The same core should work with different runtimes depending on the target environment: a Linux runtime for development and testing, and eventually more constrained runtimes for embedded devices.\r\n\r\nThis separation changes how \"lightweight\" is understood. It is not just about removing features, but about drawing clear boundaries between protocol behavior, storage, networking, scheduling, and hardware-specific execution. The first step is a Linux proof of concept: a small single-user ActivityPub server used to validate the framework before moving toward constrained runtimes.\r\n\r\nThe goal is not to present a finished server, but to share what it means to design ActivityPub as a portable engine rather than a monolithic application. Feder is still early-stage, but its direction is clear: one ActivityPub core, many runtimes — and the question of what kinds of Fediverse software become possible when federation logic can be separated from the machine it runs on.",{"title":5375,"describe":5376,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FMUHKX7",{"id":5381,"room":5382,"start":3465,"end":4830,"language":96,"track":5383,"speakers":5385,"zh":5393,"en":5396,"tags":5399,"uri":5400},"MWDT3S",{"en":39,"zh-hant":10},{"id":384,"name":5384},{"en":386,"zh-hant":10},[5386],{"id":5387,"avatar":5388,"zh":5389,"en":5392},"DABTYH","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FDABTYH_DuDfSV2.jpg",{"name":5390,"bio":5391},"York","- Blockchain Developer\r\n- Security Researcher\r\n- DefiHackLabs White-Hat",{"name":5390,"bio":5391},{"title":5394,"describe":5395,"type":30},"ERC-7984 如何讓鏈上資產徹底隱形？","在傳統 EVM 的世界裡，透明度是一把雙面刃，所有的代幣餘額、轉帳金額與交易流向都赤裸裸地攤在鏈上，這不僅讓使用者毫無隱私可言，更滋生了無數的 Mempool 搶跑攻擊 (Front-running) 與惡意 MEV。\r\n\r\n> **如果我們能讓代幣的「狀態」變成密文呢？**\r\n\r\n\r\n本議程將深入探討由 OpenZeppelin 與 Zama 共同推動的革命性草案——_**ERC-7984 (Confidential Fungible Token)**_。本議程將解析如何利用全同態加密 (FHE) 技術與指標型 (Pointer-based) 架構，讓 Smart Contract 在「不解密」的情況下，直接對加密餘額進行加減運算與驗證。本議程將從底層架構出發，對比傳統 ERC-20 的明文盲點，實地剖析 fhEVM 環境下的合約開發挑戰，並探討這項技術將如何徹底顛覆 DeFi 的資安與隱私遊戲規則。",{"title":5397,"describe":5398,"type":30},"How Does ERC-7984 Make On-Chain Assets Completely Invisible?","Public ledgers expose everything. Every token balance and transaction flow is visible to anyone, making users prime targets for MEV bots and front-running attacks. But what if smart contracts could compute on encrypted data without ever decrypting it?\r\n\r\nEnter ERC-7984 (Confidential Fungible Token). This session unpacks the power of Fully Homomorphic Encryption (FHE) and how the fhEVM ecosystem—spearheaded by OpenZeppelin and Zama—is bringing true privacy to Web3. I will break down the architectural leap from standard ERC-20s, explore the practical challenges of building encrypted smart contracts, and discuss how hiding on-chain states will fundamentally redefine DeFi security.",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FMWDT3S",{"id":5402,"room":5403,"start":4830,"end":4831,"language":96,"track":5404,"speakers":5406,"zh":5414,"en":5417,"tags":5419,"uri":5420},"NBAKRA",{"en":684,"zh-hant":10},{"id":686,"name":5405},{"en":688,"zh-hant":689},[5407],{"id":5408,"avatar":5409,"zh":5410,"en":5413},"G3HJDZ","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FG3HJDZ_Blbm7ED.png",{"name":5411,"bio":5412},"司改會數位法小組","就公權力與私領域數位化的過程中所涉及之人權議題進行研議與行動。工作範圍可二分為「具體事件型」與「整體結構型」，前者著重當前正在發生的個案或法規草案，後者則身負台灣數位人權法治之領導重任，推展長期系統性與結構性之改善，使我國在走向數位化的過程中，法制架構上不只是消極不侵害權利，亦積極有助產生更加自由的新數位社會環境。",{"name":5411,"bio":5412},{"title":5415,"describe":5416,"type":30},"數位權利基本法：一場以民間立法推進開源政策的行動實驗","本計畫源於民間司法改革基金會（司改會）的一場草根立法實踐，嘗試以社會運動的集體動能，集結科技、法律與人權背景的公民夥伴們共同草擬《數位權利基本法》。這不只是一份法典的產出，更是公民社群主動參與治理，試圖將所凝聚的價值轉化為推動政策的法條的努力過程。\r\n\r\n在這場次中，我們首先會分享《數位權利基本法》的關鍵精神與內涵，再來進一步提出法律的制定本身即是一種價值觀的優先排列選擇—而民間對於開源、commons價值的嚮往與推進，也可以嘗試以「法律制定」作為載體，還促進更全面而穩健的政策落實。\r\n\r\n《數位權利基本法》第十七條「公共程式原則」，探討如何落實公共資金開發成果開放予全民共享，落實數位公地的精神；第十八條「健全數位環境之落實」，旨在從法制面上構築支持數位公民權的基礎設施，確保群眾能在健康、永續的數位生態中共同協作與成長。\r\n\r\n這是一場進行中的嘗試，我們誠摯邀請大家針對條文設計、實務落地及倡議路徑共同腦力激盪，期盼能將這份民間立法成功轉化為具體政策推力。",{"title":5418,"describe":5416,"type":30},"Digital Bill of Rights: Driving Open Source Policy via Civic Legislation Initiatives",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FNBAKRA",{"id":5422,"room":5423,"start":3692,"end":5424,"language":96,"track":5425,"speakers":5427,"zh":5435,"en":5438,"tags":5439,"uri":5440},"NEJWDB",{"en":1253,"zh-hant":10},"2026-08-09T09:40:00+08:00",{"id":1258,"name":5426},{"en":1260,"zh-hant":1261},[5428],{"id":5429,"avatar":5430,"zh":5431,"en":5434},"HHUYMA","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FHHUYMA_3aTvNr1.webp",{"name":5432,"bio":5433},"林彥廷","林彥廷，Google DeepMind Research Scientist，專注語言模型後訓練（post-training）；開源社群貢獻者，曾創建 Taiwan-LLM 系列模型。",{"name":5432,"bio":5433},{"title":5436,"describe":5437,"type":30},"[Prime Session] From Taiwan-LLM to the Frontier: What Open Source Taught Me, and Why I Still Miss It","這場演講會分享我從 Taiwan-LLM 一路做開源 LLM 的經驗，以及換了位置走進前線研究後，回頭看開源 LLM 模型的發展。",{"title":5436,"describe":5437,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FNEJWDB",{"id":5442,"room":5443,"start":3504,"end":3505,"language":42,"track":5444,"speakers":5446,"zh":5454,"en":5457,"tags":5460,"uri":5461},"NERQJY",{"en":39,"zh-hant":10},{"id":384,"name":5445},{"en":386,"zh-hant":10},[5447],{"id":5448,"avatar":5449,"zh":5450,"en":5453},"LUXLCF","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FLUXLCF_2et8w5w.webp",{"name":5451,"bio":5452},"Albert Cheng","XueDAO - Core Contributor\r\nResearcher - Capital Layer",{"name":5451,"bio":5452},{"title":5455,"describe":5456,"type":30},"量子電腦下，Bitcoin 該何去何從？","① Bitcoin 在量子電腦下的攻擊面\r\n② 公鑰、私鑰、地址：鏈上看得到 hash ≠ 看得到 pubkey\r\n③ 為什麼「9 分鐘破解」是 BTC 的生死線？\r\n④ BIP-360, BIP 361 在做什麼？\r\n⑤ CRQC 真的來了我們怎麼盡早知道？\r\n⑥ 防範公鑰外洩 - HNDL（Harvest Now, Decrypt Later）\r\n⑦ 為什麼後量子升級總是被排到最後一步？",{"title":5458,"describe":5459,"type":30},"Bitcoin in the Quantum Era — Where Do We Go From Here?","① Bitcoin's attack surface under quantum computers\r\n  ② Public key, private key, address: seeing a hash on chain ≠ seeing the pubkey\r\n  ③ Why \"9-minute crack\" is the line in the sand for BTC\r\n  ④ What BIP-360 and BIP-361 are doing\r\n  ⑤ How do we know a CRQC has actually arrived — as early as possible?\r\n  ⑥ Defending against pubkey exposure — HNDL (Harvest Now, Decrypt Later)\r\n  ⑦ Why post-quantum upgrades always get pushed to last",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FNERQJY",{"id":5463,"room":5464,"start":4013,"end":5465,"language":70,"track":5466,"speakers":5468,"zh":5472,"en":5475,"tags":5476,"uri":5477},"NJSSSE",{"en":326,"zh-hant":10},"2026-08-09T12:15:00+08:00",{"id":3268,"name":5467},{"en":3270,"zh-hant":3271},[5469],{"id":3676,"avatar":3677,"zh":5470,"en":5471},{"name":3679,"bio":3680},{"name":3679,"bio":3680},{"title":5473,"describe":5474,"type":30},"Hybrid AI: The Next Pattern for AI-Powered Apps","Most applications treat AI as a cloud-only feature — send a request, wait for a response, pay per token. Open models like Gemma, Llama, DeepSeek, and Phi freed developers from proprietary APIs, but not from costs: you still need to host them somewhere — GPU servers, infrastructure, per-request pricing. You can run models directly on the user's device — zero inference costs, offline access, full data privacy. But there's a trade-off: models small enough to run on a phone or in a browser can't handle every task.\r\nHybrid AI is the answer to this dilemma. Combine cloud and on-device models in a single application: simple and privacy-sensitive tasks run on-device, complex reasoning and multimodal workloads go to the cloud. The problem is that you end up maintaining two completely different inference stacks: different runtimes (TFLite, LiteRT-LM, llama.cpp), different model formats (TFLite flatbuffers vs GGUF), different quantization strategies, and different streaming APIs — all within one app.\r\nIn this talk, I'll show how to solve this problem with Genkit — an open-source AI framework from Google with an open plugin system. Each plugin adapts a specific runtime or model format to a unified pipeline: genkit_flutter_gemma for TFLite\u002FLiteRT models, genkit_llamadart for GGUF. Engine-specific details are hidden behind a single API for flows, structured output, tool calling, and agentic workflows. Need a new runtime? Write a plugin. New model format? Same story. On-device inference runs across Android, iOS, macOS, Windows, Linux, and Web — switching between cloud and local execution is a one-line model reference change.\r\nWe'll explore hybrid patterns in practice: how to route between on-device and cloud based on model capabilities, connectivity, and task complexity. We'll cover the real trade-offs — quantization impact on output quality, memory constraints across platforms, cold start latency, and when hybrid actually makes sense versus pure cloud or pure edge.\r\nOpen-source models made on-device AI possible. Open-source tooling makes hybrid AI extensible.",{"title":5473,"describe":5474,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FNJSSSE",{"id":5479,"room":5480,"start":3568,"end":3548,"language":96,"track":5481,"speakers":5483,"zh":5491,"en":5494,"tags":5497,"uri":5498},"NLCRV9",{"en":326,"zh-hant":10},{"id":3268,"name":5482},{"en":3270,"zh-hant":3271},[5484],{"id":5485,"avatar":5486,"zh":5487,"en":5490},"RFCLRX","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FRFCLRX_ZHBCOjH.jpg",{"name":5488,"bio":5489},"徐方繹 Fngi Shiu","Fngi 專注於 AI 安全治理與 LLM 紅隊測試。曾任多家台灣機構 CISO，同時擔任 GDG Taipei Organizer 及台灣金融研訓院 AI 安全講師。擁有超過十年資安與全端工程經驗。",{"name":5488,"bio":5489},{"title":5492,"describe":5493,"type":30},"你的 LLM App 上線第一天就該做的三件事（但你可能一件都沒做）","你花了一個週末串好 RAG、接上 MCP、部署了一個會自動回 Issue 的 LLM Agent——然後你就直接上線了。沒有 input validation、沒有權限控制、System Prompt 裸奔。別擔心，大家都這樣，而且不需要資安背景也能跟上。今天我們直接開 terminal，打給你看。\r\n這場議程會告訴你上線前該做好的三件事：一、管好你的 System Prompt——我們會現場用 Prompt Injection 把它整段偷出來，然後示範怎麼用 NemoClaw Guardrail 擋住。二、清洗你的 RAG 資料來源——我們會在知識庫文件裡埋 Indirect Prompt Injection，讓你的 Agent 做出它不該做的事，然後示範 content sanitization 怎麼實作。三、限制你的 Agent 權限——我們會讓一個有過多權限的 Agent 自己刪掉自己的資料，然後用 least privilege 原則加上 tool call 權限控制修好它。\r\n每一件事都是「現場用 OpenClaw 打給你看，再用 NemoClaw 修給你看」，全部開源，git clone 就能跑。你會帶走一份可以直接丟進 repo 的安全 checklist。",{"title":5495,"describe":5496,"type":30},"Three Things Your LLM App Should Have Done on Day One (But Probably Didn't)","You spent a weekend wiring up RAG, connecting MCP, and deploying an LLM Agent that auto-replies to GitHub Issues — then shipped it straight to production. No input validation, no permission controls, system prompt wide open. Don't worry, everyone does it. No security background needed to follow along — we'll open a terminal and show you live.\r\nThis talk covers three things you should do before going live: First, protect your system prompt — we'll steal it with prompt injection, then block the attack with NemoClaw Guardrail. Second, sanitize your RAG data sources — we'll plant an indirect prompt injection in a knowledge base document and make your Agent do things it shouldn't, then show how content sanitization works. Third, limit your Agent's permissions — we'll let an over-privileged Agent delete its own data, then fix it with least-privilege principles and tool call permission controls.\r\nEvery attack uses OpenClaw, every fix uses NemoClaw — all open source, git clone and go. You'll leave with a security checklist ready to drop into your repo.",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FNLCRV9",{"id":5500,"room":5501,"start":3668,"end":4013,"language":96,"track":5502,"speakers":5504,"zh":5512,"en":5515,"tags":5518,"uri":5519},"NMZWGM",{"en":407,"zh-hant":10},{"id":410,"name":5503},{"en":412,"zh-hant":412},[5505],{"id":5506,"avatar":5507,"zh":5508,"en":5511},"3MZGML","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002F3MZGML_wEN2Bhx.webp",{"name":5509,"bio":5510},"David Peng","在聯發科打工的前端工程師，喜歡寫 Rust ，曾在 JSDC 分享，有在經營 substack 歡迎去看看\r\n\r\nFrontend engineer working at MediaTek. Rust enthusiast. Previously spoke at JSDC. Also runs a Substack, feel free to check it out.",{"name":5509,"bio":5510},{"title":5513,"describe":5514,"type":30},"自架 centrifugo 實作即時通知機制","實作伺服器端推播，直覺第一個想到的通常是 WebSocket，但自行維護 WebSocket 並不簡單：你需要處理 ping\u002Fpong 心跳、連線管理、斷線重連，在微服務架構下還要決定由哪個服務負責維護這些狀態\r\n\r\n我們在生產環境中使用 Centrifugo 將近一年，用來處理檔案上傳後的即時進度通知。這場分享會介紹我們如何導入這套開源工具，讓後端只需呼叫一支 RESTful API 就能推播訊息給對應的客戶端，不再需要自己管理 WebSocket 連線\r\n\r\n內容涵蓋：\r\n- Centrifugo 的基本概念與 channel 機制\r\n- Server 端設定與啟動\r\n- 後端整合方式（以 RESTful API 推播）\r\n- 前端 SDK 連線與訂閱\r\n- 實際導入的心得與注意事項",{"title":5516,"describe":5517,"type":30},"Self-Hosting Centrifugo for Real-Time Push Notifications","When implementing server-side push, WebSocket is usually the first thing that comes to mind — but maintaining WebSocket connections yourself is far from simple. You need to handle ping\u002Fpong heartbeats, connection management, and reconnection logic. In a microservices architecture, deciding which service owns those connections adds yet another layer of complexity.\r\n\r\nWe've been running Centrifugo in production for nearly a year, using it to deliver real-time progress notifications for file upload workflows. This talk covers how we integrated this open-source tool so that our backend only needs to call a single RESTful API to push messages to clients — no WebSocket state management required.\r\n\r\nTopics covered:\r\n- Centrifugo's core concepts and channel model\r\n- Server setup and getting started\r\n- Backend integration via RESTful API publish\r\n- Frontend SDK connection and subscription\r\n- Practical lessons learned from production",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FNMZWGM",{"id":5521,"room":5522,"start":4639,"end":4640,"language":96,"track":5523,"speakers":5525,"zh":5533,"en":5536,"tags":5539,"uri":5540},"NUPSVF",{"en":326,"zh-hant":10},{"id":3268,"name":5524},{"en":3270,"zh-hant":3271},[5526],{"id":5527,"avatar":5528,"zh":5529,"en":5532},"SUVVXB","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FSUVVXB_O9vuwAP.webp",{"name":5530,"bio":5531},"Kent Huang","Software Engineer @ Recce 與 AI 搏鬥中，嘗試不要被淹沒在 AI 洪流的工程師一枚",{"name":5530,"bio":5531},{"title":5534,"describe":5535,"type":30},"AI 也要期中考？ The Story Behind Building a Benchmark for Agent Workflow Development","要考期中考？AI 沒過也會有這天。開發 Agent Workflow 時要怎麼分辨。表現變好是 Model 的功勞，還是好的 Workflow 帶來的功勞？這時就需要一套好的驗證方法：期中考！但是道高一尺，魔高一丈。你想得到的，AI 也想得到。上網作弊、偷看答案，只有你想不到，沒有 AI 做不到。本節的故事背景，是我們在開發開源 Agent Workflow 時，設計內部 benchmark 所碰到的各種大小事。\r\n\r\nHow do you evaluate the performance of an agent workflow? When things get better, is it the model doing the work, or the workflow itself? A good benchmark is like a midterm exam. It tells you who actually studied.\r\nBut AI is smarter than you expect. It will cheat by searching the web, peek at other answers, and find every loophole you forgot to close. AI always finds a way to surprise you.\r\nIn this session, we'll share the story behind building a benchmark for an open-source agent workflow. Every twist, every workaround, and every time the AI quietly outsmarted our test design.",{"title":5537,"describe":5538,"type":30},"The Story Behind Building a Benchmark for Agent Workflow Development","How do you evaluate the performance of an agent workflow? When things get better, is it the model doing the work, or the workflow itself? A good benchmark is like a midterm exam. It tells you who actually studied.\r\nBut AI is smarter than you expect. It will cheat by searching the web, peek at other answers, and find every loophole you forgot to close. AI always finds a way to surprise you.\r\nIn this session, we'll share the story behind building a benchmark for an open-source agent workflow. Every twist, every workaround, and every time the AI quietly outsmarted our test design.",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FNUPSVF",{"id":5542,"room":5543,"start":3968,"end":3969,"language":328,"track":5544,"speakers":5546,"zh":5550,"en":5553,"tags":5554,"uri":5555},"PAQLLM",{"en":3731,"zh-hant":10},{"id":3734,"name":5545},{"en":3736,"zh-hant":3736},[5547],{"id":1566,"avatar":1567,"zh":5548,"en":5549},{"name":1569,"bio":1570},{"name":1569,"bio":1570},{"title":5551,"describe":5552,"type":30},"Unified database provisioning and management on Kubernetes","Running production-grade databases on Kubernetes is becoming increasingly common, but managing their lifecycle remains fragmented and complex for SRE and DevOps teams. Critical operations such as scaling, monitoring, backup, and restore currently require navigating distinct, database-specific APIs and tools. This complexity prevents teams from fully realizing the operational efficiency and uniformity that Kubernetes provides.\r\n\r\nThis talk introduces OpenEverest, a CNCF Sandbox project designed to address this operational gap. OpenEverest provides a single, unified UI and CLI to manage SRE functions for popular open-source databases such as PostgreSQL and MySQL deployed on Kubernetes. It abstracts away database-specific differences, offering standardized control for diverse databases.\r\n\r\nJoin us to learn how OpenEverest simplifies the path to production readiness, reduces operational toil, and is building a pioneering open-source database management layer.",{"title":5551,"describe":5552,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FPAQLLM",{"id":5557,"room":5558,"start":3754,"end":3755,"language":42,"track":5559,"speakers":5561,"zh":5569,"en":5572,"tags":5575,"uri":5576},"PBK3Y7",{"en":39,"zh-hant":10},{"id":384,"name":5560},{"en":386,"zh-hant":10},[5562],{"id":5563,"avatar":5564,"zh":5565,"en":5568},"7MTLSS","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002F7MTLSS_aoct9RJ.webp",{"name":5566,"bio":5567},"Oskar (oskarth)","Oskar Thorén 目前在以太坊基金會共同領導 IPTF（Institutional Privacy Task Force，機構隱私工作小組），專注於協助機構導入 Ethereum，並滿足其隱私需求。他近期曾擔任 PSE 的策略顧問。此前，他建立了 mopro，一個讓行動裝置上的客戶端證明（client-side proving）更容易實作的工具。他也曾創立 Vac.dev 並擔任研究總監；Vac.dev 是一個研發組織，致力於為去中心化網路建構公共利益協議，重點領域包括隱私與通訊。\r\n\r\n網站：oskarth.com\r\nTwitter：@oskarth\r\n\r\nOskar Thorén is co-leading IPTF (Institutional Privacy Task Force) at the Ethereum Foundation, focused on onboarding institutions onto Ethereum and meeting their privacy requirements. He recently served as a strategic advisor at PSE. Previously, he created mopro, a tool for making client-side proving on mobile easier. He also founded and was Director of Research at Vac.dev, an R&D organization building public good protocols for the decentralized web, with a focus on privacy and communication. Website: oskarth.com Twitter: @oskarthx",{"name":5566,"bio":5567},{"title":5570,"describe":5571,"type":30},"ZK Made Easy：零知識證明入門","零知識證明（Zero-Knowledge Proofs, ZK）讓我們可以「證明一件事是真的」，但不需要公開背後的敏感資料。它正在成為隱私保護、區塊鏈擴容、數位身分與可驗證運算的重要基礎技術，而且許多關鍵工具與協定都在開源社群中發展。\r\n\r\n這場分享會用非技術背景也能理解的方式，介紹 ZK 的核心直覺與入門技術概念：什麼是 proof、prover、verifier、witness、public input，ZK 如何把一個問題轉成 circuit 與 constraints，以及 completeness、soundness、zero knowledge 這些基本性質代表什麼。\r\n\r\n我們會用簡單例子，例如證明自己超過 18 歲、Where's Waldo 與 Sudoku，說明「證明知道答案」和「不公開答案」如何同時成立。接著會看 ZK 在今天的實際應用，包括私密交易、匿名投票、ZK rollups、身分證明、選擇性揭露與可驗證運算。\r\n\r\n本分享不需要密碼學或數學背景。聽眾離開時，應該能用簡單語言解釋 ZK，理解基本技術組成，判斷哪些問題適合用 ZK 解決，並知道如何透過開源文章、工具、專案與社群開始學習、實驗或貢獻。",{"title":5573,"describe":5574,"type":30},"ZK Made Easy: An Introduction to Zero-Knowledge Proofs","Zero-knowledge proofs let us prove that something is true without revealing the sensitive data behind it. They are becoming important infrastructure for privacy, blockchain scaling, digital identity, and verifiable computation, and many key tools and protocols are being developed in open-source communities.\r\n\r\nThis beginner-friendly talk introduces both the intuition and the basic technical vocabulary of ZK: proofs, provers, verifiers, witnesses, public inputs, circuits, constraints, and the core properties of completeness, soundness, and zero knowledge.\r\n\r\nWe will use simple examples such as proving you are over 18, Where's Waldo, and Sudoku to explain how someone can prove they know something without revealing the underlying secret. We will then look at real-world applications including private payments, anonymous voting, ZK rollups, identity, selective disclosure, and verifiable computation.\r\n\r\nNo cryptography or math background is required. After the talk, the audience should be able to explain ZK in simple terms, understand its basic technical components, recognize what kinds of problems ZK can help solve, and know where to start learning, experimenting, or contributing through open-source articles, tools, projects, and communities.",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FPBK3Y7",{"id":5578,"room":5579,"start":3548,"end":3464,"language":96,"track":5580,"speakers":5582,"zh":5586,"en":5589,"tags":5592,"uri":5593},"PMJV7D",{"en":67,"zh-hant":10},{"id":3993,"name":5581},{"en":3995,"zh-hant":3995},[5583],{"id":436,"avatar":437,"zh":5584,"en":5585},{"name":439,"bio":440},{"name":439,"bio":440},{"title":5587,"describe":5588,"type":30},"我的 360 野生街景遊記","把 360 相機架到車頂上（或者頭頂上），然後一起來走遍大街小巷吧！\r\n我的 360 街景遊記——怎麼拍、如何處理、上傳到哪裡、資料怎麼 Open。\r\n一起來扮演野生街景車，記錄下你的國內外遊蹤。",{"title":5590,"describe":5591,"type":30},"My 360 Street View Travelogue","Mount a 360 camera on your car roof—or even on your head—and let’s explore the streets together.\r\n\r\nThis is my 360° street-view travel story: how I shoot, process, publish, and open the data. I’ll walk through the full workflow and show you how to turn your travels into immersive, open street-level experiences.\r\n\r\nJoin and become the wild street view car, recording your journeys across cities, countries, and everywhere in between.",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FPMJV7D",{"id":5595,"room":5596,"start":3643,"end":3549,"language":96,"track":5597,"speakers":5599,"zh":5606,"en":5609,"tags":5610,"uri":5611},"PPF3YH",{"en":167,"zh-hant":10},{"id":3526,"name":5598},{"en":3528,"zh-hant":3528},[5600,5603],{"id":5330,"avatar":5331,"zh":5601,"en":5602},{"name":5333,"bio":5334},{"name":5333,"bio":5334},{"id":4411,"avatar":4412,"zh":5604,"en":5605},{"name":4414,"bio":4415},{"name":4414,"bio":4415},{"title":5607,"describe":5608,"type":30},"PDF 解封之後—一起動手打造 Classifier 吧","為什麼要開這個議題？\r\n* **資料分類是模型進化的階梯—** 延續先前的 FinePDFs，如何讓這些碎片化的文字變成可被 LLM 吸收的「領域專長」，關鍵就在於高品質的分類與標納。\r\n* **訓練 LLM 的重要步驟—** 訓練或微調模型前，你需要一個精準的分類器來過濾雜訊與分類資料！\r\n* **運用高效率的 Few-Shot 方法打造分類器—** 分類器的訓練方法百百種，本議程預計使用 SetFit ，可以在極少量的標註樣本下，快速打造出強大的分類器。\r\n* **評估與分析—** 不只談訓練，更需要進行評估與分析，科學地驗證模型究竟是真的還是瞎猜。\r\n\r\n誰適合來聽這個議程？\r\n* **開源狂熱份子—** 想了解開源 LLM 的技術細節。\r\n* **LLM 領域的初學者—** 想知道在「Prompt Engineering」之外，如何透過微調技術（Fine-tuning）改善資料流程的開發者。\r\n\r\n這個議題期望帶來什麼樣的影響力？\r\n* **我們是 TwinkleAI—** 一個致力於建立「臺灣在地資料集與模型」的社群，也藉由讀書會等活動，推廣 LLM 相關知識的普及，目前已公開多筆資料集以及數顆繁體中文模型。\r\n* **人人都能學會的 LLM—** 透過大量公開的資料，人人都能入門 LLM 的相關知識；並且理解 LLM 的運作脈絡，你可以更有效率的馴服 AI！！",{"title":5607,"describe":5608,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FPPF3YH",{"id":5613,"room":5614,"start":4536,"end":4054,"language":70,"track":5615,"speakers":5617,"zh":5625,"en":5628,"tags":5629,"uri":5630},"PSC9EM",{"en":237,"zh-hant":10},{"id":241,"name":5616},{"en":243,"zh-hant":243},[5618],{"id":5619,"avatar":5620,"zh":5621,"en":5624},"UNTT7A","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FUNTT7A_2zcMCVy.webp",{"name":5622,"bio":5623},"Pawel Lisewski @palekiwi","Based in Taiwan. I appreciate the beauty of competing paradigms although deep at heart I am a functional programmer who naively loves Haskell and Nix and - less naively - Rust.\r\nI think it is fun to be a polyglot in both the machine and human world so try asking me a question in any language and there's a non-zero chance I'll manage to answer ;)",{"name":5622,"bio":5623},{"title":5626,"describe":5627,"type":30},"The Deterministic Rubyist: Taming Your AI Pair Programmer with Nix","AI coding agents are the new \"high-speed interns\" of the Ruby world. They can be eager and fast, but leave them unconstrained and you risk wrecking your system. Tie them up too tight and they'll end up hallucinating instead of producing results.\r\n\r\nThe answer isn't more documentation or prompt engineering; it’s a better contract.\r\n\r\nIn this talk, we’ll explore how Nix - the functional, deterministic package manager - serves as a bridge between human developers and AI agents. We will move beyond the \"it works on my machine\" era into the \"it works in the contract\" era.\r\n\r\nWe'll discuss how a Nix Flake acts as a shared mental model: a declarative, cryptographic receipt of every dependency, compiler, and system library your project requires.\r\n\r\nWe will dive into:\r\n\r\n* The Bundler Analogy: Translating Nix concepts into a language Rubyists already speak (Gemfile → Flake, Lockfile → Lock).\r\n* Effectiveness through Shared Context: How to use Nix to ensure your AI agent has the exact tools it needs (like ruby-lsp or specific C-extensions)\r\n* Security as an Enabler: Using Nix in tandem with Docker to create \"principled sandboxes.\" We’ll look at real-world patterns from my opensource AI sandboxing tool.\r\n* The \"Double Gemfile\" Pattern: A practical tip for managing dev-only gems and agent-specific tools within a Nix-based workflow.\r\n\r\nBy the end of this session, you’ll see that Nix is a powerful tool for the modern Rubyist to build secure, reproducible, and highly effective environments where humans and AI can collaborate without the chaos.",{"title":5626,"describe":5627,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FPSC9EM",{"id":5632,"room":5633,"start":3548,"end":3549,"language":42,"track":5634,"speakers":5636,"zh":5640,"en":5643,"tags":5644,"uri":5645},"PSMBUE",{"en":684,"zh-hant":10},{"id":686,"name":5635},{"en":688,"zh-hant":689},[5637],{"id":2156,"avatar":2157,"zh":5638,"en":5639},{"name":2159,"bio":2160},{"name":2159,"bio":2160},{"title":5641,"describe":5642,"type":30},"開盒資訊公務員：投入政府服務的資訊人員在想什麼呢？","世界各地都有針對政府機關開源的倡議，比如公共資金、公共程式。但是在實務上，資訊職系的公務員在做資訊採購案時，必須因應來自督導長官、執行廠商、以及平行單位的壓力，因此符合在地脈絡的政策路徑便非常重要。開放、公開、透明、協作的智慧政府服務該怎麼達成？在有標準答案之前，我們先來聽聽公務員怎麼說。",{"title":5641,"describe":5642,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FPSMBUE",{"id":5647,"room":5648,"start":3667,"end":3668,"language":70,"track":5649,"speakers":5651,"zh":5659,"en":5662,"tags":5663,"uri":5664},"QALAL8",{"en":326,"zh-hant":10},{"id":3268,"name":5650},{"en":3270,"zh-hant":3271},[5652],{"id":5653,"avatar":5654,"zh":5655,"en":5658},"JUMDFL","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FJUMDFL_9j50avB.webp",{"name":5656,"bio":5657},"Teo Richie","Richie is a Lead Experience Engineer based in Kuala Lumpur, Malaysia, with five years of experience building software products. He works at the intersection of backend systems and user-facing product, with a focus on making complex infrastructure feel simple to use. His recent work includes Alfred, a multi-agent orchestration platform. This project pushed him deep into the practical challenges of running AI systems in production, which is what this talk is about.",{"name":5656,"bio":5657},{"title":5660,"describe":5661,"type":30},"Wiring AI Agents to the Real World: The Integration Layer Nobody Talks About","Everyone talks about prompts, models, and RAG. Nobody talks about the 80% of work that actually gets AI agents running in production. \r\n\r\nThis talk is about the integration layer. The OAuth flows, channel routing, credential management, scheduling infrastructure, and observability tooling that separates a working demo from a system your users can rely on every day. \r\n\r\nI'll draw from a system I've participated in building: Alfred, a multi-agent orchestration platform. We'll start with integrations. An agent is only as useful as what it can reach. Connecting to Gmail, Slack, Google Calendar, Microsoft 365, or Telegram sounds straightforward until you're dealing with token expiry, scope changes, and multiple users with different credentials. We'll look at how to build a credential layer that handles all of this cleanly, so agents can act on behalf of users without the whole thing falling apart when a token rotates. \r\n\r\nFrom there we'll talk about routing. In a real deployment, messages come in from multiple channels simultaneously. Telegram, Slack, and web chat all have different UX contracts and different audiences. The answer isn't to write channel-aware agent code. It's to build a routing layer that maps channels to agents declaratively, so you can change behavior without touching your agent logic. \r\n\r\nWe'll also cover scheduling. Agents are far more powerful when they run on a schedule rather than waiting to be asked. We'll look at how to treat cron jobs as first-class agent invocations, with full run history and cost tracking so you know exactly what every scheduled run is doing and what it's spending. \r\n\r\nFinally we'll talk about observability. If you can't see what your agent is doing in real time, and you can't track what it's costing you, you can't trust it in production. We'll look at patterns for tracking subagent spawns, monitoring delegation health, and surfacing live activity so problems surface before your users notice them. \r\n\r\nBy the end of this talk you'll have a clear mental model of the integration layer and a set of concrete, framework-agnostic patterns you can take back to whatever stack you're building on.",{"title":5660,"describe":5661,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FQALAL8",{"id":5666,"room":5667,"start":3708,"end":3709,"language":42,"track":5668,"speakers":5670,"zh":5674,"en":5677,"tags":5678,"uri":5679},"QNHYZ9",{"en":9,"zh-hant":10},{"id":3645,"name":5669},{"en":3647,"zh-hant":3648},[5671],{"id":3332,"avatar":3333,"zh":5672,"en":5673},{"name":3335,"bio":3336},{"name":3335,"bio":3336},{"title":5675,"describe":5676,"type":30},"(Compiler) Homun-lang: A Rust-Based Scripting Language for AI-Native Developing","Ever dreamed of creating your own programming language? This talk walks through the real journey — from building a Mermaid-to-ASCII compiler as a warm-up, to designing   Homun, a language that transpiles to Rust. You'll learn the compiler pipeline (tokenizer → parser → AST → codegen), why transpiling beats writing a full compiler, and the hard design lessons: auto-detection traps, syntax iteration pain, and why hemi-self-hosting keeps things sane. Practical takeaways for anyone curious about language design in the AI era.",{"title":5675,"describe":5676,"type":30},[402],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FQNHYZ9",{"id":5681,"room":5682,"start":3567,"end":3568,"language":1469,"track":5683,"speakers":5685,"zh":5693,"en":5696,"tags":5697,"uri":5698},"QPNZNG",{"en":67,"zh-hant":10},{"id":3993,"name":5684},{"en":3995,"zh-hant":3995},[5686],{"id":5687,"avatar":5688,"zh":5689,"en":5692},"QP7KLA","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FQP7KLA_5D0aPKb.jpg",{"name":5690,"bio":5691},"Hiroshi Miura","Lead developer of the OmegaT project. Thirty years of experience in free, libre, and open-source software development, including contributions to the Linux kernel, Samba, and taking the maintainer of the py7zr Python library, aqtinstall utility, and more.\r\nAlso, the representative director of OpenStreetMap Foundation Japan, and editor and translator of WeeklyOSM.",{"name":5690,"bio":5691},{"title":5694,"describe":5695,"type":30},"Join the Open Map Community: FOSS4G, State of the Map Asia, and Beyond","The open mapping community has its own conference culture — and this summer, two big events happen right after COSCUP, just across the sea.\r\n\r\nIn this session, Hiroshi Miura — representative director of OpenStreetMap Foundation Japan and editor of weeklyOSM — will introduce these events and invite you to join.\r\n\r\nFOSS4G Hiroshima (September 1–3, conference + September 4–5, sprint): The global conference for open-source geospatial technology. This year it comes to Hiroshima, Japan. Thousands of GIS developers, researchers, and map lovers gather together. Stay for the sprint and hack with the community!\r\n\r\nState of the Map Asia 2026 (September 6–7, Osaka): The regional OSM community gathering for Asia. Contributors from across the continent share their work and meet each other in person.\r\nCome to COSCUP in Taipei, then head to Hiroshima for FOSS4G, and finish in Osaka for SotM Asia. An open-source summer trip across the region!\r\n\r\nAnd after the events — write about your experience! weeklyOSM is a global OSM media, run by volunteers, that collects and shares community news from around the world every week. Your blog post or trip report can reach the whole OSM community worldwide.\r\n\r\nNo special knowledge needed. If you have ever edited a map, used open geodata, or just wondered how the map on your phone gets made — this session is for you.",{"title":5694,"describe":5695,"type":30},[],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FQPNZNG",{"id":5700,"room":5701,"start":3504,"end":3505,"language":96,"track":5702,"speakers":5704,"zh":5718,"en":5721,"tags":5722,"uri":5723},"QRE9V9",{"en":192,"zh-hant":10},{"id":196,"name":5703},{"en":198,"zh-hant":198},[5705,5712],{"id":5706,"avatar":5707,"zh":5708,"en":5711},"NYAHUJ","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FNYAHUJ_StWPDbN.webp",{"name":5709,"bio":5710},"EricccTaiwan (周呈陽)","Familiar with the C & Operating system.\r\nLearning Linux Kernel Internals and contributing `sched_ext` currently.\r\nAnd I love hitting the gym.\r\n\r\n- GitHub : [EricccTaiwan](https:\u002F\u002Fgithub.com\u002FEricccTaiwan)\r\n- [Kernel contribution](https:\u002F\u002Fgit.kernel.org\u002Fpub\u002Fscm\u002Flinux\u002Fkernel\u002Fgit\u002Ftj\u002Fsched_ext.git\u002Flog\u002F?qt=grep&q=Cheng-Yang+Chou&h=for-next)\r\n- [LinkedIn](https:\u002F\u002Fwww.linkedin.com\u002Fin\u002Fyphbchou0911\u002F)\r\n\r\nI want to be a **Full-time Kernel Hacker**!!",{"name":5709,"bio":5710},{"id":5713,"avatar":5714,"zh":5715,"en":5717},"NGWBDW","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FNGWBDW_C8IlerN.png",{"name":5716,"bio":10},"邱柏穎",{"name":5716,"bio":10},{"title":5719,"describe":5720,"type":30},"Learn and Play with sched_ext","Linux **v6.12** 引入的 `sched_ext` (scx) 允許開發者藉由 eBPF，在使用者空間動態載入或抽換 CPU 排程器。\r\n本議程預計探討以下：\r\n\r\n- `sched_ext` and `eBPF`\r\n- 剖析 **sched_ext** 在 Kernel 如何實作 ( `kernel\u002Fsched\u002Fext*`)\r\n- 回顧 **BORE** scheduler\r\n- Play with **scx_rustland_core**\r\n- 以 eBPF 實作 BORE scheduler [scx_BORE]",{"title":5719,"describe":5720,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FQRE9V9",{"id":5725,"room":5726,"start":3754,"end":3755,"language":328,"track":5727,"speakers":5729,"zh":5737,"en":5740,"tags":5741,"uri":5742},"RDNPEW",{"en":9,"zh-hant":10},{"id":3645,"name":5728},{"en":3647,"zh-hant":3648},[5730],{"id":5731,"avatar":5732,"zh":5733,"en":5736},"LEZHVZ","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FLEZHVZ_J9QwWqN.webp",{"name":5734,"bio":5735},"Boen","A software engineer interested in anything Turing-complete.",{"name":5734,"bio":5735},{"title":5738,"describe":5739,"type":30},"UXN - Virtual Machine Design in a Single C File","[UXN](https:\u002F\u002F100r.co\u002Fsite\u002Fuxn.html) is a stack-based virtual machine designed by the [Hundred Rabbits](https:\u002F\u002F100r.co\u002Fsite\u002Fabout.html). It's small enough to fit in [a single C source file](https:\u002F\u002Fwiki.xxiivv.com\u002Fetc\u002Fuxnmin.c.txt), and simple enough to be ported across a wide range of platforms, including GBA, Nintendo DS, DOS, PlayStation Vita, Raspberry Pi, and many more.\r\n\r\nUXN is minimal but not limited. Paired with the [Varvara](https:\u002F\u002Fwiki.xxiivv.com\u002Fsite\u002Fvarvara.html) devices, it can support games, drawing tools, editors, and even [a graphical operating system](https:\u002F\u002Fwiki.xxiivv.com\u002Fsite\u002Fpotato.html?utm_source=chatgpt.com).\r\n\r\nUXN is simple and hackable, which makes it a useful way to learn the fundamentals of virtual machine design. This session will walk through the [specification](https:\u002F\u002Fwiki.xxiivv.com\u002Fsite\u002Fuxntal.html?utm_source=chatgpt.com), including opcodes, memory layout, and device I\u002FO, and then examine the reference implementation, [uxn2](https:\u002F\u002Fgit.sr.ht\u002F~rabbits\u002Fuxn2).",{"title":5738,"describe":5739,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FRDNPEW",{"id":5744,"room":5745,"start":3568,"end":3548,"language":96,"track":5746,"speakers":5748,"zh":5756,"en":5759,"tags":5762,"uri":5763},"RES9MM",{"en":192,"zh-hant":10},{"id":196,"name":5747},{"en":198,"zh-hant":198},[5749],{"id":5750,"avatar":5751,"zh":5752,"en":5755},"ZWCFQD","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FZWCFQD_Ha1g7Ye.webp",{"name":5753,"bio":5754},"徐彥翔（c8763yee）","就讀中正大學資訊工程研究所(OSLAB)\r\n\r\nI Use {Arch,Fedora} BTW",{"name":5753,"bio":5754},{"title":5757,"describe":5758,"type":30},"從個人電腦到資料中心：Linux 記憶體管理的演化","本議程將從 Linux 虛擬記憶體 (VM) 子系統的演化脈絡出發，分三個部分介紹：\r\n\r\n1. RMAP 發展史: 從 Linux v2.5 的 pte_chain（per-page reverse mapping）出發，探討其 scalability 問題；接著介紹 Andrea Arcangeli 在 2004 年導入 anon_vma 的設計背景，以及如何回應 VM scalability、分析 anon_vma_chain 如何改善大量 fork workload 的擴展性問題；最後介紹以 interval tree 改善大型 address space 下的 VMA lookup 與 RMAP walk scalability\r\n2. RMAP 在現代頁面置換中的作用: 介紹 Multi-Gen LRU（MGLRU）如何透過 generation-based aging，減少 reclaim 過程中對 page table 與 reverse mapping 的重複掃描，並說明其在現代資料中心 workload 下的重要性。\r\n3. 實驗量化 RMAP Overhead: 使用 bpftrace 掛載 kprobe:rmap_walk，量測 Transparent HugePage (THP) 啟用與停用時的 latency 分布差異，並分析 huge page split、folio reclaim 與 deferred split queue 等路徑如何影響 reverse mapping 成本。",{"title":5760,"describe":5761,"type":30},"from COMPUTER to COMPUTing centER: the history of linux memory management","The presentation is structured into three main segments:\r\n\r\n1. The History of RMAP\r\nWe begin by tracing the origins of RMAP from the pte_chain (per-page reverse mapping) in Linux v2.5, examining its initial scalability challenges. The talk will explore the design background of anon_vma, introduced by Andrea Arcangeli in 2004 to address VM scalability, and discuss how anon_vma_chain mitigates scalability bottlenecks in heavy fork workloads. Finally, we will cover the adoption of interval trees to optimize VMA lookups and RMAP walk scalability within extremely large address spaces.\r\n\r\n2. The Role of RMAP in Modern Page Replacement\r\nThe second segment highlights the integration of RMAP within contemporary memory management strategies, particularly Multi-Gen LRU (MGLRU). We will discuss how MGLRU utilizes generation-based aging to significantly reduce redundant scanning of page tables and reverse mappings during the page reclaim process, emphasizing its critical importance in optimizing modern datacenter workloads.\r\n\r\n3. Experimental Quantification of RMAP Overhead\r\nIn the final section, we present an empirical analysis of RMAP overhead. By leveraging bpftrace to hook into kprobe:rmap_walk, we measure and compare the latency distributions when Transparent HugePage (THP) is enabled versus disabled. Furthermore, we will analyze how specific kernel paths(such as huge page split, folio reclaim, and deferred split queues) impact the overall cost of reverse mapping operations in real-world scenarios.",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FRES9MM",{"id":5765,"room":5766,"start":3465,"end":4830,"language":96,"track":5767,"speakers":5769,"zh":5777,"en":5780,"tags":5783,"uri":5784},"RFU7JA",{"en":282,"zh-hant":10},{"id":3670,"name":5768},{"en":3672,"zh-hant":3673},[5770],{"id":5771,"avatar":5772,"zh":5773,"en":5776},"8DF8EJ","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002F8DF8EJ_oFRzfKT.jpg",{"name":5774,"bio":5775},"Kevin","Cloud AI GDE",{"name":5774,"bio":5775},{"title":5778,"describe":5779,"type":30},"用Antigravity 2.0 \u002FIDE 打造你的Telegram 小助手","用Antigravity 2.0 \u002FIDE 打造你的 Telegram 智慧小助手 (Telegram Assistant Bot) - 5 Labs Series\r\nEnglish Summary below\r\n\r\n本專案是一個循序漸進的 Telegram 機器人（智慧小助手）開發教學與範例專案。專案共劃分為 5 個獨立可執行的 Lab 目錄，展示如何從零建置一個整合 Google Gemini 3.5 Flash 與 Imagen 3 的智慧助理。\r\n\r\n📂 專案目錄結構 (Lab 規劃)\r\n本專案採用功能遞增的方式，每個 Lab 都是一個完整且可獨立執行的專案：\r\n\r\n📂 Lab 1: 基礎 Bot 框架\r\n建置 Telegram Bot 連線、基本指令（\u002Fstart 與 \u002Fhelp）與 Echo 訊息回覆。\r\n📂 Lab 2: AI 簡報生成器\r\n串接 Gemini 3.5 Flash 生成結構化大綱，並使用 python-pptx 自動繪製簡報（支援 Light\u002FDark 主題設定）。\r\n📂 Lab 3: AI 圖片生成器\r\n串接 Gemini Imagen 3 圖片生成模型，提供 \u002Fdraw [比例] \u003C描述> 指令，支援 1:1、16:9、9:16、4:3、3:4 等多種長寬比。\r\n📂 Lab 4: AI 圖像修改器\r\n上傳圖片至機器人，機器人會進入等待指令狀態，呼叫 Gemini 多模態與 Imagen 3 協同為您修改原圖並回傳。\r\n📂 Lab 5: 個人隨手筆記本 (最終版)\r\n引入本機 SQLite 資料庫儲存個人筆記，並實作 \u002Faddnote 與 \u002Fnotes。清單底部提供內聯按鈕（Inline Keyboard）非同步、無縫刪除個別筆記。此 Lab 為功能最完整的終極版本。",{"title":5781,"describe":5782,"type":30},"Build your Telegram assistant with Antigravity 2.0 \u002FIDE","Build Your Telegram Assistant Bot with Antigravity 2.0 \u002FIDE - 5 Labs Series\r\n\r\nEnglish Summary below\r\n\r\nThis project is a step-by-step tutorial and example project for developing a Telegram bot (smart assistant). The project is divided into 5 independent executable Lab directories, demonstrating how to build a smart assistant integrating Google Gemini 3.5 Flash and Imagen 3 from scratch.\r\n\r\n📂 Project Directory Structure (Lab Planning)\r\n\r\nThis project adopts an incremental approach, with each Lab being a complete and independently executable project:\r\n\r\n📂 Lab 1: Basic Bot Framework\r\nBuilding Telegram bot connectivity, basic commands (\u002Fstart and \u002Fhelp), and Echo message replies.\r\n\r\n📂 Lab 2: AI Presentation Generator\r\nConnects to Gemini 3.5 Flash to generate structured outlines and uses python-pptx to automatically draw presentations (supports Light\u002FDark theme settings).\r\n\r\n📂 Lab 3: AI Image Generator\r\nConnects to the Gemini Imagen 3 image generation model, providing the `\u002Fdraw [scale] \u003Cdescription>` command, supporting various aspect ratios such as 1:1, 16:9, 9:16, 4:3, and 3:4.\r\n\r\n📂 Lab 4: AI Image Editor\r\nUpload an image to the bot, which will enter a waiting state for instructions, calling on Gemini multimodal and Imagen 3 to collaboratively modify the original image and send it back.\r\n\r\n📂 Lab 5: Personal Notebook (Final Version)\r\nIntegrates a local SQLite database to store personal notes and implements `\u002Faddnote` and `\u002Fnotes`. An inline keyboard at the bottom of the list allows for asynchronous, seamless deletion of individual notes. This Lab is the most complete and ultimate version.",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FRFU7JA",{"id":5786,"room":5787,"start":3777,"end":3887,"language":328,"track":5788,"speakers":5790,"zh":5798,"en":5801,"tags":5802,"uri":5803},"RJW7QF",{"en":1253,"zh-hant":10},{"id":1258,"name":5789},{"en":1260,"zh-hant":1261},[5791],{"id":5792,"avatar":5793,"zh":5794,"en":5797},"Q9BUNH","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FQ9BUNH_DPz1DlS.webp",{"name":5795,"bio":5796},"Shu Muto","Shu Muto was a maintainer for Kubernetes Dashboard and several OpenStack Dashboard plugins since 2015. He organizes Kubernetes Upstream Training Japan to increase contributors and promotes open-source culture as an organizer of Cloud Native Community Japan. He is committed to fostering innovation and collaboration in the open-source community as a CNCF Ambassador, a TODO OSPO Ambassador and a OSPO Alliance Contributor.\r\n\r\nhttps:\u002F\u002Fsessionize.com\u002Fshu-muto\u002F",{"name":5795,"bio":5796},{"title":5799,"describe":5800,"type":30},"[Prime Session] Building Bridges from Asia to Global Open Source","Open source is not only software we use, but a way for engineers, students, companies, communities, and even countries to learn and grow together. However, many people in Asia still face barriers when trying to join global open source communities: language, cultural distance, lack of confidence, and limited organizational support.\r\n\r\nIn this keynote, I will share how we have approached these challenges in Japan through Kubernetes Upstream Training Japan, Cloud Native Community Japan, localization activities, and corporate open source strategy. These efforts started as local-language onboarding, but gradually evolved into a pathway that helps users take their first steps toward real upstream contribution.\r\n\r\nI will also discuss why sustainable contribution cannot depend only on heroic individuals. Companies, universities, OSPOs, policy makers, foundations, and local communities all have roles to play in building participation infrastructure. Through Japan’s experience, I hope to explore how Asian communities can learn from each other and build stronger bridges to global open source.",{"title":5799,"describe":5800,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FRJW7QF",{"id":5805,"room":5806,"start":3487,"end":4908,"language":70,"track":5807,"speakers":5809,"zh":5817,"en":5820,"tags":5823,"uri":5824},"RUFNVU",{"en":93,"zh-hant":10},{"id":3467,"name":5808},{"en":3469,"zh-hant":3469},[5810],{"id":5811,"avatar":5812,"zh":5813,"en":5816},"TSBW9B","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FTSBW9B_KM6E4VP.jpg",{"name":5814,"bio":5815},"Jerry, Jiancong Zhao","Jerry Zhao is the Chief SDV Architect in the R&D division of Panasonic Automotive Systems Co., Ltd., with extensive development experience in cutting-edge technologies for software-defined vehicles (SDVs), including virtualization, cloud native, and digital twins. He is also deeply engaged in open-source and automotive community initiatives. Jerry leads the SDV Expert Group within Automotive Grade Linux (AGL) and serves as the APAC Regional Hub Lead for SOAFEE—two of the most active organizations promoting open collaboration in the SDV ecosystem.\r\nJerry Zhao現任日本松下汽車電子系統有限公司研發部門的軟體定義車輛（SDV）首席架構師，擁有豐富的前沿技術開發經驗，涵蓋虛擬化、雲原生技術及數位分身等領域。他亦積極參與開源與汽車產業社群活動，目前領導 Automotive Grade Linux（AGL）中的 SDV 專家小組，並擔任 SOAFEE 的亞太區區域樞紐負責人。這兩個組織皆為推動 SDV 領域開放合作最活躍的社群之一。",{"name":5814,"bio":5815},{"title":5818,"describe":5819,"type":30},"Software Defined Vehicles: Innovations Through Open Source Collaboration","Software Defined Vehicles (SDVs) mark a significant evolution in the automotive industry, shifting the focus from traditionally hardware-centric designs to software-driven systems. Earlier, automotive systems were built using a hardware-first approach; today, vehicles are increasingly powered by complex, software-based solutions.\r\n\r\nSeveral open-source alliances—such as AGL, ELISA, SOAFEE, Eclipse, COVESA, Autosar, and others—are uniting to accelerate the development of SDVs. In this session, we will provide an overview of these initiatives, explaining how these alliances are shaping the future of automotive technology.\r\n\r\nIn particular, the AGL SDV Expert Group is at the forefront of this movement, committed to developing common SDV platforms through open-source collaboration. As the leader of this group, the speaker will share the group’s vision, roadmap, and detailed activities, offering a closer look at AGL’s efforts in the SDV space.\r\n\r\nThis presentation is designed to initiate collaboration between the automotive open-source community and COSCUP attendees, and will be conducted in both English and Chinese.",{"title":5821,"describe":5822,"type":30},"軟體定義車輛（SDV）：透過開源協作推動創新","軟體定義車輛（Software‑Defined Vehicles, SDVs）正引領汽車產業的重大轉型，車輛架構逐漸從以硬體為核心，轉向由軟體所定義與差異化。過去汽車系統多圍繞固定硬體元件設計，而現代車輛則越來越依賴可更新、具彈性且高度軟體化的平台。\r\n為了加速此一轉型，多個開放原始碼聯盟──如 AGL、ELISA、SOAFEE、Eclipse、COVESA、AUTOSAR 等──正跨越 SDV 生態系的不同層面展開協作。本場演講將概述這些組織的角色與方向，並說明其如何共同推動下一世代汽車平台的發展。\r\n本次分享旨在促進汽車開源社群與 COSCUP 參與者之間的交流與合作，並將以英語及中文進行。",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FRUFNVU",{"id":5826,"room":5827,"start":3465,"end":4830,"language":70,"track":5828,"speakers":5830,"zh":5838,"en":5841,"tags":5844,"uri":5845},"S7KAQQ",{"en":121,"zh-hant":10},{"id":4034,"name":5829},{"en":4036,"zh-hant":4036},[5831],{"id":5832,"avatar":5833,"zh":5834,"en":5837},"ZYUW93","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FZYUW93_xMDrvNH.webp",{"name":5835,"bio":5836},"Jim DeLaHunt","Jim DeLaHunt is a Canadian software engineer with a long career in technology which intertwines with human language and culture. He has long experience in type, text layout, font technology, Unicode, and internationalisation. \r\n\r\n- He worked in the Type group at Adobe Systems, where he helped ship Japanese OpenType fonts, and led development of a system for typesetting \"gaiji\" (外字) in Japanese typesetting. \r\n- He was a solo software engineering consultant, doing back-end development and project management for industries as varied as electric power distribution and long-haul trucking.\r\n- He has been a frequent presenter at the Internationalisation and Unicode Conferences in North America.\r\n- He is a member of the Canadian delegation to the ISO 10646 standard, which parallels The Unicode Standard.\r\n- He has volunteered with numerous free libre and internationalisation projects, including Joomla, Wikipedia, IMUG, MacPorts, and MusicBrainz. He frequents Fediverse conferences like FediForum, FediCon, and recently SWICG.\r\n\r\nYou can find him at @jdlh@mstdn.ca, at jdlh.com, and on the Fediverse wherever the #i18n or #GloballyInclusive hashtags appear.",{"name":5835,"bio":5836},{"title":5839,"describe":5840,"type":30},"@小明@範例.測試, or, Fediverse handles in every language","A person's handle is their name and address on the Fediverse. But despite people using the Fediverse in many languages from all over the world, almost all handles are limited to Latin script letters and numbers: **@somebody@example.com** . Instead, people should be able to use a handle in their own language and characters, e.g. **@小明@範例.測試** or **@김@예시.테스트**. We call this \"globally inclusive Fediverse handles\".\r\n\r\nThis talk explains the reasons why handles are now limited to letters and numbers. We look at the role of specifications, and existing software such as Mastodon. We look at the history of internationalising other address types such as email addresses ( जिम@डाटामेल.भारत ) and domain names ( http:\u002F\u002Fยูเอทดสอบ.ไทย ). We list some rules from DNS, Unicode, and IETF about what characters should be allowed and forbidden: the Fediverse can build on them. \r\n\r\nWe propose a path forward, based on the history of email addresses and domain names. That path starts with brave pioneers, who want to use their own language in their Fediverse handles, even when it is difficult. The next step is helping extend specifications so they are globally inclusive. Third is contributing to open source software to remove their limitations for handle characters. Hopefully, many of those pioneers are at COSCUP. Maybe this meeting will be the time when we meet and start to work together.",{"title":5842,"describe":5843,"type":30},"[Chinese translation of title to follow if proposal is accepted]","[Chinese translation of abstract to follow if proposal is accepted]",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FS7KAQQ",{"id":5847,"room":5848,"start":4258,"end":4259,"language":96,"track":5849,"speakers":5851,"zh":5859,"en":5862,"tags":5863,"uri":5864},"S8GEJP",{"en":1253,"zh-hant":10},{"id":901,"name":5850},{"en":903,"zh-hant":904},[5852],{"id":5853,"avatar":5854,"zh":5855,"en":5858},"BDKCQQ","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FBDKCQQ_oaj4U6r.jpg",{"name":5856,"bio":5857},"王佑中","從 1992 年開始參與開放原始碼運動。開發 Linux 上第一套中文的終端機及相關的中文化套件。隨後至 UC Irvine 攻讀博士。參與 real-time Linux 的開發。過去二十年多半從事嵌入式系統的開發工作。目前任教長庚大學人工智慧學系，為長庚 AI 中心副主任。",{"name":5856,"bio":5857},{"title":5860,"describe":5861,"type":30},"makeslide: NotebookLM 不夠好用？那就自己做一個語音簡報生成工具","NotebookLM 是非常有用的 AI 學習與內容轉換工具，但在語言、流程控制、輸入格式、在地化需求與可客製化程度上，仍有不少限制。與其等待平台功能逐步完善，不如嘗試用開源方式打造一個更符合特定場景的替代工具。makeslide 是一個語音簡報生成與播放系統，目標是讓使用者可以從 PDF 或簡報大綱出發，自動產生可播放的語音簡報。系統目前可接受 PDF，將頁面轉為圖片並抽取文字；也可以接受以 ## Slide XX 分隔的簡報大綱，進一步產生投影片圖片、逐字稿與語音內容。目前專案以繁體中文為主要支援語言，後端採 Fastify 與 TypeScript，前端使用 React、Vite 與 Tailwind，並提供上傳、處理、查詢與播放所需的 API 與處理管線。\r\n\r\n本議程將分享 makeslide 的設計動機、系統架構、PDF 到語音簡報的處理流程，以及開發過程中對 NotebookLM-like 工具「可自建、可修改、可在地化」的實作經驗。",{"title":5860,"describe":5861,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FS8GEJP",{"id":5866,"room":5867,"start":4140,"end":3643,"language":1469,"track":5868,"speakers":5870,"zh":5878,"en":5881,"tags":5882,"uri":5883},"SAMSGH",{"en":1063,"zh-hant":10},{"id":1065,"name":5869},{"en":1067,"zh-hant":1068},[5871],{"id":5872,"avatar":5873,"zh":5874,"en":5877},"CBLLBE","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FCBLLBE_M1HqaTH.webp",{"name":5875,"bio":5876},"Kozo Nishida","[Kozo Nishida is a Technical Scientist at RIKEN](https:\u002F\u002Fwww.linkedin.com\u002Fin\u002Fkozonishida\u002F), where he works at the intersection of life sciences and open source software. His work focuses on improving the interoperability, reproducibility, and accessibility of computational research, particularly in metabolomics and bioinformatics.\r\n\r\nWith over a decade of experience supporting scientific research environments, Kozo has contributed to the development and maintenance of research software and data standards. He is actively involved in international open science communities, including the [Bioconductor](https:\u002F\u002Fwww.bioconductor.org\u002F) community and global working groups around data standards such as [mzTab-M](https:\u002F\u002Fgithub.com\u002FHUPO-PSI\u002FmzTab-M). His recent work explores how to connect experimental data formats with downstream analysis and visualization tools to enable more seamless scientific workflows.\r\n\r\nKozo is also deeply engaged in community building and education. He has been involved in organizing events such as [PyData meetups in Japan](https:\u002F\u002Fosaka.pydata.org\u002F) and [community-driven workshops](https:\u002F\u002Fluma.com\u002Fclaudecommunity), and he contributes to initiatives that promote inclusive participation in open source. As part of his work with multilingual efforts in open science communities, he has supported the translation and localization of technical training materials, helping make them accessible to a broader audience.\r\n\r\nIn addition to his contributions to research software and community activities, Kozo is exploring the use of AI tools to enhance developer productivity and knowledge sharing. His recent efforts include designing agent-based workflows for translating structured technical documentation while preserving its integrity—bridging the gap between automation and human-centered open source practices.\r\n\r\nThrough his work, Kozo aims to lower barriers to participation in open science and to foster collaboration across languages, disciplines, and communities.",{"name":5875,"bio":5876},{"title":5879,"describe":5880,"type":30},"Translating Open Source Lessons with AI (Without Breaking Them)","Open source educational materials often come with their own “dialects” of Markdown. For example, [The Carpentries Workbench](https:\u002F\u002Fcarpentries.github.io\u002Fworkbench\u002F) extends Markdown with special directives, metadata blocks, and conventions that are essential for building lessons. These are not just formatting details—they are part of the executable structure of the content.\r\n\r\nToday, AI tools make it easy to translate documentation quickly. You can generate a full draft in seconds. But if you’ve tried this on structured teaching materials, you may have seen what happens next: broken builds, modified syntax, or translations that ignore the intent of the lesson design. AI is powerful, but it doesn’t automatically understand which parts of a document must never be touched.\r\n\r\nIn this talk, I’ll share a practical approach to solving this problem using “[agent skills](https:\u002F\u002Fagentskills.io\u002Fhome)” designed specifically for translating structured technical content. These skills act as guardrails: they protect non-translatable syntax while allowing AI to focus on natural language parts. The goal is not just speed, but safe and maintainable translation workflows.\r\n\r\nI’ll present a real-world example from the [Bioconductor](https:\u002F\u002Fwww.bioconductor.org\u002F) community, where we applied this approach to translate Carpentries-style training materials. With these agent skills, we were able to significantly accelerate translation while keeping lessons fully buildable and consistent.\r\n\r\nMore importantly, this is not just a tooling story—it’s a community story. These translated materials are now being used in workshops across different regions, enabling participants to learn in their own languages. Translation becomes more than a task; it becomes an entry point for contributing to open source. You don’t need to start with code—you can start by helping your local community access knowledge.\r\n\r\nAt COSCUP, where community and openness are at the center, this talk aims to share both practical techniques and a perspective: AI can amplify community efforts, but only if we design workflows that respect how open source projects actually work.\r\n\r\nIf you care about documentation, localization, or making open source more inclusive, this talk is for you.",{"title":5879,"describe":5880,"type":30},[],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FSAMSGH",{"id":5885,"room":5886,"start":3823,"end":3590,"language":328,"track":5887,"speakers":5889,"zh":5893,"en":5896,"tags":5897,"uri":5898},"SC3REC",{"en":922,"zh-hant":10},{"id":330,"name":5888},{"en":332,"zh-hant":332},[5890],{"id":4385,"avatar":4386,"zh":5891,"en":5892},{"name":4388,"bio":4389},{"name":4388,"bio":4389},{"title":5894,"describe":5895,"type":59},"Self-Hosting 101: A real-time self-hosting workshop","This is a continuation from my talk that I gave at UbuCon Asia 2025: https:\u002F\u002Fevents.canonical.com\u002Fevent\u002F127\u002Fabstracts\u002F838\u002F\r\n\r\nThis time, I want to host a workshop where I'll walk people through the steps on how to setup Ubuntu server from scratch for self-hosting. I'll start with setting up an LXD VM with Ubuntu server. I'll walk through securing the server using firewall, fail2ban and other things. I'll then cover setting up a reverse-proxy like nginx and then host a simple web application on it. If time permits, I can also cover setting up podman for containers. I can also go a bit deep into the concepts and cover the basics of routing and tunneling as well.\r\n\r\nRequirements from attendees:\r\n\r\n- Laptop.\r\n\r\n- It would be great if they could setup LXD on their device (I'll walk people through this if most of them aren't familiar with LXD.)\r\n\r\n- It would be great if they could pre-download the Ubuntu server ISO (https:\u002F\u002Fubuntu.com\u002Fdownload\u002Fserver).\r\n\r\nMore info: https:\u002F\u002Fevents.canonical.com\u002Fevent\u002F146\u002Fcontributions\u002F945\u002F",{"title":5894,"describe":5895,"type":59},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FSC3REC",{"id":5900,"room":5901,"start":3505,"end":3732,"language":96,"track":5902,"speakers":5904,"zh":5908,"en":5911,"tags":5914,"uri":5915},"SRLDEB",{"en":684,"zh-hant":10},{"id":686,"name":5903},{"en":688,"zh-hant":689},[5905],{"id":714,"avatar":715,"zh":5906,"en":5907},{"name":717,"bio":718},{"name":717,"bio":718},{"title":5909,"describe":5910,"type":30},"網路域名基礎建設背後的自由與開源軟體","雖然網際網路乘載了目前生活中從電子郵件、社群媒體到人工智慧系統的各種服務，但許多網際網路的使用者對於支持整個網際網路系統的基礎建設，也就是域名（domain name）系統的瞭解並不深入，逞論理解域名系統中的各種自由與開源軟體的存在。這些仰賴社群參與與投入來維護的自由與開源軟體，也在網路基礎建設中扮演重要的角色，不過由於其重要性經常被忽略，以至於政府與國際組織在思考網路基礎建設的相關政策時，往往無法制定適當的政策來協助社群發展。\r\n\r\n在 2025 年 10 月，網際網路名稱與數位位址分配機構（ICANN）下的網路安全與穩定諮詢委員會（SSAC）發布了 SAC 132 號報告，報告主題是《在自由與開源軟體上運行的網路域名系統》（The Domain Name System Runs on Free and Open Source Software (FOSS)）。本議程希望從這份報告出發，帶領參與者瞭解域名系統以及與域名系統相關的自由與開源軟體服務；並且更進一步探討若要制定政策支持這些開源社群與貢獻者，有哪些考量應該在政策設計的過程中被納入。並且盤點台灣目前的政策是否符合本報告提出的建議。\r\n\r\n本議程也將一併介紹 ICANN 組織、會議與台灣的相關參與機會，包含由台灣網路資訊中心支持的 TWNIC Academy 活動，以及相關的參與經驗分享。",{"title":5912,"describe":5913,"type":30},"An Introduction to ICANN SAC 132: The Domain Name System Runs on Free and Open Source Software (FOSS)","Although the internet underpins a wide range of services in modern life—from email and social media to artificial intelligence systems—many users have only a limited understanding of the infrastructure that supports it, particularly the Domain Name System (DNS). Even fewer are aware of the role played by free and open source software within the DNS ecosystem. These community-driven FOSS projects are essential to the functioning of internet infrastructure, yet their importance is often overlooked. As a result, governments and international organizations frequently struggle to design effective policies that adequately support the communities maintaining these systems.\r\nIn October 2025, the Internet Corporation for Assigned Names and Numbers’s Security and Stability Advisory Committee released Report SAC132, titled The Domain Name System Runs on Free and Open Source Software (FOSS). Building on this report, this session aims to help participants understand the DNS and the FOSS services that support it. It will further explore key considerations for policymakers seeking to support these open-source communities and contributors, and assess whether Taiwan’s current policies align with the recommendations outlined in the report.\r\nThe session will also introduce Internet Corporation for Assigned Names and Numbers as an organization, its meeting structure, and opportunities for participation from Taiwan. This includes programs supported by the Taiwan Network Information Center, such as the TWNIC Academy, along with reflections and experiences from past participants.",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FSRLDEB",{"id":5917,"room":5918,"start":5919,"end":4639,"language":96,"track":5920,"speakers":5922,"zh":5930,"en":5933,"tags":5935,"uri":5936},"SSN9Y7",{"en":326,"zh-hant":10},"2026-08-09T09:35:00+08:00",{"id":3268,"name":5921},{"en":3270,"zh-hant":3271},[5923],{"id":5924,"avatar":5925,"zh":5926,"en":5929},"H9PJPV","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002Foie_ZBmmBLx7r7zm_QNFhfvF.jpg",{"name":5927,"bio":5928},"Che Chia Chang","Che-Chia Chang 是一名專注於後端開發、開發維運、容器化應用及 Kubernetes 開發與管理的技術專家，同時也是 Microsoft 最有價值專業人士（MVP）。\r\n\r\n活躍於台灣技術社群，經常在 CNTUG、DevOps Taipei、GDG Taipei、Golang Taipei Meetup 等社群分享 DevOps、SRE、Kubernetes 及雲端運算相關技術。致力於推動開發與維運的最佳實踐，並熱衷於研究與應用最新的雲端與 AI 技術。\r\n\r\n個人部落格：https:\u002F\u002Fchechia.net\r\n\r\nChe-Chia Chang is a technology expert specializing in backend development, DevOps, site reliability engineering (SRE), containerized applications, and Kubernetes development and management. He is also recognized as a Microsoft Most Valuable Professional (MVP).\r\n\r\nActively engaged in the Taiwanese tech community, he frequently shares insights on DevOps, SRE, Kubernetes, and cloud computing at CNTUG, DevOps Taipei, GDG Taipei, and Golang Taipei Meetup. Passionate about promoting best practices in development and operations, he continuously explores and applies the latest advancements in cloud and AI technologies.\r\n\r\nhttps:\u002F\u002Fchechia.net",{"name":5927,"bio":5928},{"title":5931,"describe":5932,"type":30},"Observe and evaluate LLM Coding Agents","在導入 LLM Coding Agent 到日常開發後，很多團隊會遇到問題：品質不穩、風險難估、模型升級缺少客觀依據。本分享會從實務流程出發，介紹如何觀測 Agent 行為、把評估流程自動化，並建立可持續迭代的 decision system。\r\n\r\n1. 為什麼 LLM Coding Agent 需要 O11y 與 Evaluation\r\n1. 從 trace 開始：觀測 agent 任務、工具調用與失敗模式\r\n1. Observability != Decision System：看得到不代表能決策\r\n1. LLM-as-a-judge 在 coding 任務中的價值與限制\r\n1. 從 daily coding logs 抽取 dataset\r\n1. 建立 regression evaluation pipeline 與 quality gate\r\n1. 用量化結果比較不同模型與 agent framework\r\n1. 把模型升級與框架切換變成可驗證工程流程",{"title":5931,"describe":5934,"type":30},"LLM coding agents are changing software delivery, but most teams still lack a reliable way to measure quality and risk before deployment. This talk shows how to move from raw observability to practical decision-making by combining tracing, dataset extraction, LLM-as-a-judge, and regression evaluation, so model and framework upgrades become testable engineering decisions instead of guesswork.",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FSSN9Y7",{"id":5938,"room":5939,"start":4055,"end":5940,"language":96,"track":5941,"speakers":5943,"zh":5947,"en":5950,"tags":5953,"uri":5954},"SUBB9G",{"en":407,"zh-hant":10},"2026-08-09T12:20:00+08:00",{"id":410,"name":5942},{"en":412,"zh-hant":412},[5944],{"id":4872,"avatar":4873,"zh":5945,"en":5946},{"name":4875,"bio":4876},{"name":4875,"bio":4876},{"title":5948,"describe":5949,"type":30},"把遊戲塞進 13KB：js13kGames 的極限壓縮實戰","js13kGames 是一個 JavaScript 遊戲開發挑戰，參賽作品必須在 zip 後 13KB 以內完成。這個限制讓遊戲開發變得很不一樣：一張圖片可能就超過限制，因此玩法設計、素材表現、程式架構、build pipeline 與壓縮策略，都必須一起思考。\r\n\r\n本議程將分享參與 js13kGames 的實務經驗，說明如何在極小容量中完成可玩的 Web 遊戲。內容會從遊戲發想、範圍控制、技術選型、素材處理、壓縮流程與體驗調整等面向，整理在 13KB 限制下常見的設計與工程取捨。\r\n\r\n除了壓縮技巧，本分享也會討論容量限制如何反過來影響遊戲設計：哪些機制值得留下、哪些功能必須刪除、如何讓操作更直覺、如何在有限空間中保留足夠的回饋、變化與 polish。",{"title":5951,"describe":5952,"type":30},"Fitting a Game into 13KB: Constraints, Compression, and Design Trade-offs in js13kGames","js13kGames is a JavaScript game development challenge where each entry must fit within 13KB after being zipped. This constraint changes the development process significantly: a single image can already exceed the limit, so gameplay design, asset representation, code structure, build pipeline, and compression strategy all need to be considered together.\r\n\r\nThis talk shares practical experience from participating in js13kGames and discusses how to build a playable Web game within an extremely small size budget. It covers game ideation, scope control, technical choices, asset handling, compression workflow, and experience refinement under the 13KB constraint.\r\n\r\nBeyond compression techniques, this talk also explores how size limitations can influence game design itself: how to control feature scope, remove content when necessary, improve control intuitiveness, and still preserve feedback, variation, and polish within limited space. Through these experiences, the talk shows how size constraints affect design decisions, engineering implementation, and the final player experience—not just how to make files smaller.",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FSUBB9G",{"id":5956,"room":5957,"start":3567,"end":4831,"language":96,"track":5958,"speakers":5960,"zh":5966,"en":5969,"tags":5972,"uri":5973},"TBDXKG",{"en":1007,"zh-hant":10},{"id":901,"name":5959},{"en":903,"zh-hant":904},[5961],{"id":5962,"avatar":3,"zh":5963,"en":5965},"RABYPP",{"name":5964,"bio":10},"William Mou",{"name":5964,"bio":10},{"title":5967,"describe":5968,"type":59},"體驗火箭導控 Balloon Popping Challenge","小時候是不是也曾幻想過寫程式控制火箭升空？現在，你的 code 真的能主宰火箭的飛行軌跡！隨著臺灣入軌火箭計畫的推進，我們正徵招「導航、制導與控制（GNC）」的跨領域軟體人才，參加國家太空中心（TASA）與陽明交通大學前瞻火箭研究中心（ARRC）共同推動的「[火箭導控軟體競賽（Balloon Popping Challenge）](https:\u002F\u002Fwww.tasa.org.tw\u002Fen-US\u002Fevents\u002Fdetail\u002F273c7841-ce66-400a-a501-1bf0f7fdc17e)」。\r\n\r\n這是個結合 [Gymnasium](https:\u002F\u002Fgithub.com\u002FFarama-Foundation\u002FGymnasium) 訓練環境與 [ActiveRocketPy](https:\u002F\u002Fgithub.com\u002FARRC-Rocket\u002FBalloonPoppingChallenge) 模擬系統(基於開源專案 [RocketPy](https:\u002F\u002Fgithub.com\u002FRocketPy-Team\u002FRocketPy)) 的火箭導控競賽。活動中會手把手帶領與會者嘗試控制火箭的推力向量控制（TVC）系統，並進一步探討如何應對陣風、感測器雜訊等複雜的真實情境。\r\n\r\n「Be a User -> Be a Contributor」是這場工作坊的核心精神！我們非常鼓勵大家參與開源貢獻：無論你想發揮 Computer Science 背景幫忙打造飛行軌跡的資料視覺化工具，還是對物理學有熱忱，想發 PR 改善火箭 Python 模組的模擬特性，這裡都有你大展身手的舞台。\r\n\r\n本工作坊無需事先報名，歡迎大家在 COSCUP 期間直接過來現場跟我們研究火箭，一起成為太空開源專案的貢獻者！如果玩出了興趣，還可以順便組隊報名參加 8 月 31 日截止的[火箭導控軟體競賽](https:\u002F\u002Fdocs.google.com\u002Fforms\u002Fd\u002Fe\u002F1FAIpQLSegCqnI4t-R_6Nxtbkf-XJ-V3L5-_DlyDxmSU_FY2Qa1lvLXQ\u002Fviewform)挑戰看看！",{"title":5970,"describe":5971,"type":59},"BalloonPoppingChallenge: Open-source Rocket GNC Simulation Software Workshop","Did you ever dream of writing code to control a rocket launch when you were a kid? Now, your code can actually dictate a rocket's flight trajectory! As Taiwan's orbital launch vehicle program advances, we are recruiting cross-disciplinary software talent in \"Guidance, Navigation, and Control (GNC)\" to participate in the \"[Balloon Popping Challenge](https:\u002F\u002Fwww.tasa.org.tw\u002Fen-US\u002Fevents\u002Fdetail\u002F273c7841-ce66-400a-a501-1bf0f7fdc17e)\" co-hosted by the Taiwan Space Agency (TASA) and the Advanced Rocket Research Center (ARRC).\r\n\r\nThis is a rocket GNC competition that integrates the [Gymnasium](https:\u002F\u002Fgithub.com\u002FFarama-Foundation\u002FGymnasium) training environment with the [ActiveRocketPy](https:\u002F\u002Fgithub.com\u002FARRC-Rocket\u002FBalloonPoppingChallenge) simulation system (based on the open-source project [RocketPy](https:\u002F\u002Fgithub.com\u002FRocketPy-Team\u002FRocketPy)). During the workshop, we will provide a hands-on guide to help attendees try controlling the rocket's Thrust Vector Control (TVC) system, and further explore how to tackle complex, real-world scenarios like wind gusts and sensor noise.\r\n\r\n\"Be a User -> Be a Contributor\" is the core spirit of this workshop! We highly encourage everyone to participate in open-source contributions: whether you want to leverage your Computer Science background to build flight trajectory data visualization tools, or you are passionate about physics and want to submit PRs to improve the simulation aspects of the core Python modules, there is a stage for you here to showcase your skills.\r\n\r\nNo prior registration is required for this workshop. We welcome everyone to drop by our session during COSCUP, explore rockets with us, and become contributors to a space open-source project! If you find it interesting, you can even team up and register for the [Rocket GNC Software Competition](https:\u002F\u002Fdocs.google.com\u002Fforms\u002Fd\u002Fe\u002F1FAIpQLSegCqnI4t-R_6Nxtbkf-XJ-V3L5-_DlyDxmSU_FY2Qa1lvLXQ\u002Fviewform) (deadline: August 31st) to take on the challenge!",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FTBDXKG",{"id":5975,"room":5976,"start":3549,"end":3776,"language":96,"track":5977,"speakers":5979,"zh":5987,"en":5990,"tags":5993,"uri":5994},"TCKVDD",{"en":407,"zh-hant":10},{"id":410,"name":5978},{"en":412,"zh-hant":412},[5980],{"id":5981,"avatar":5982,"zh":5983,"en":5986},"HCSQVN","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FHCSQVN_45pbskO.webp",{"name":5984,"bio":5985},"IlyaL","Vue.js Taiwan 成員\r\nVueUse 核心團隊成員\r\nUnplugin 外部合作貢獻者",{"name":5984,"bio":5985},{"title":5988,"describe":5989,"type":30},"你早已身處 UnJS 宇宙，只是還不知道","世界是由原子組成的：微小、單純、各司其職，卻能堆疊出萬物。你有沒有想過，同樣的概念也能用來打造軟體？\r\n\r\n這正是 UnJS 的核心哲學。UnJS 是由 Nuxt 作者 Pooya Parsa 發起的開源生態系，遵循 Unix 哲學，專注打造單一用途、高品質、且不綁定框架的 JavaScript 工具，例如 h3、Nitro、ofetch、unbuild、unstorage 等。每個套件就像一顆原子，把一件事做到極好，組合起來便撐起了 Nuxt、Vite 周邊，乃至無數現代 JS 專案的基礎建設。\r\n\r\n這場 talk 會帶你認識 UnJS 這個組織、它的設計理念與運作方式，並走訪生態系中幾個關鍵的「原子」，看看它們各自解決什麼問題。\r\n\r\n如果你對這個浩瀚的宇宙也有興趣，歡迎一起來參加這場星際冒險，讓我帶你一起航向 UnJS 宇宙！",{"title":5991,"describe":5992,"type":30},"You're Already in the UnJS Universe, You Just Don't Know It Yet","The world is built from atoms: tiny, simple, each doing one job, yet together they form everything. Have you ever thought the same idea could be used to build software?\r\n\r\nThat's the core philosophy of UnJS, an open-source ecosystem started by Nuxt's creator Pooya Parsa. Following the Unix philosophy, UnJS builds single-purpose, high-quality, framework-agnostic JavaScript tools such as h3, Nitro, ofetch, unbuild, and unstorage. Each package is like an atom that does one thing exceptionally well, and together they power Nuxt, much of the Vite ecosystem, and countless modern JavaScript projects.\r\n\r\nThis talk introduces UnJS as an organization, its design philosophy, and how it operates, and walks through a few key \"atoms\" in the ecosystem to see what problems each one solves.\r\n\r\nIf this vast universe sounds interesting to you, come join the adventure and let me take you on a journey into the UnJS universe！",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FTCKVDD",{"id":5996,"room":5997,"start":3567,"end":3568,"language":70,"track":5998,"speakers":6000,"zh":6008,"en":6011,"tags":6012,"uri":6013},"TJBUBN",{"en":1253,"zh-hant":10},{"id":901,"name":5999},{"en":903,"zh-hant":904},[6001],{"id":6002,"avatar":6003,"zh":6004,"en":6007},"AZEBPT","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FAZEBPT_ztGKR8U.webp",{"name":6005,"bio":6006},"Alfonso Cancellara","Alfonso Cancellara is a Technical Account Manager at **Red Hat** specialized in OpenShift and the cloud-native ecosystem, though his **true passion lies in the \"Open Source Way\"**, the art of balancing enterprise-grade stability with the rapid, unbridled innovation of the community.\r\n\r\nA familiar face at Red Hat summits and IBM events, Alfonso is equally at home in local community meetups. He is a **dedicated community builder**, notably serving as the primary orchestrator for Red Hat’s involvement in Linux Day Italy.",{"name":6005,"bio":6006},{"title":6009,"describe":6010,"type":30},"Beyond the Code: Defining Open Source AI","The tech industry is currently witnessing a linguistic land grab. **As AI floods the ecosystem, the definition of \"Open Source\" is being stretched to its breaking point**. While traditional Open Source Software relies on deterministic code and transparent licenses, the non-linear, opaque nature of Large Language Models challenges these old paradigms. Without a rigorous redefinition of openness, we risk a future of black-box innovation.\r\n\r\nTo bridge this gap, we will start by establishing a baseline understanding of AI mechanics. With this core understanding established, we will deconstruct the AI stack to understand **why traditional software licenses fail**. We’ll move beyond the surface level to explore:\r\n- **The Pull Request Paradox**: Why modifying a model’s behavior is fundamentally different from a standard pull request and what that means for community contribution.\r\n- **Sovereignty vs. Access**: A deep dive into the \"Openness Spectrum\", comparing different Open Source AI standards, including the Open Source AI Definition by the Open Source Initiative.\r\n- **The Anatomy of Openwashing**: How providers use the \"Open\" label to gain community trust while gatekeeping the curation of training data and the compute recipes essential for true Open Source AI.\r\n\r\nAttendees will walk away with a **comprehensive framework for evaluating AI \"openness\"**, a technical understanding of why \"open weights\" does not equal \"open source\", and the vocabulary needed to advocate for true digital sovereignty in their communities or organizations.",{"title":6009,"describe":6010,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FTJBUBN",{"id":6015,"room":6016,"start":3567,"end":3568,"language":70,"track":6017,"speakers":6019,"zh":6027,"en":6030,"tags":6031,"uri":6032},"TK3BKR",{"en":1063,"zh-hant":10},{"id":1065,"name":6018},{"en":1067,"zh-hant":1068},[6020],{"id":6021,"avatar":6022,"zh":6023,"en":6026},"FYDRSP","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FFYDRSP_lI7AvLJ.webp",{"name":6024,"bio":6025},"Masahiko Hashimoto","Masahiko Hashimoto is an open-source software (OSS) enthusiast with a deep passion for Linux and programming. An avid user of openSUSE, he writes articles exploring \"slightly unusual ways\" to use it. He self-publishes these articles as doujinshi (fanzines) and distributes them at events such as Comic Market.\r\nHe is also an active member of the \"Tokaido LUG\" (Linux User Group) community. He frequently attends offline meetups held all over Japan, where he enjoys engaging in lively discussions about Linux with fellow members.\r\n\r\nProfessionally, He works in AI development, focusing specifically on \"Federated Learning\"—a machine learning technique that trains AI models across networks. As a result of his research, he has developed a federated learning framework and released it as OSS. In the future, he aims to build a system where an open-source Kana-Kanji conversion engine (IME) continuously evolves every day through federated learning.",{"name":6024,"bio":6025},{"title":6028,"describe":6029,"type":30},"Yuino: OSS kana-kanji conversion system that combines LLM and a dictionary-based approach","Yuino is an OSS Kana-Kanji conversion system that integrates LLMs with a dictionary-based approach.\r\n\r\nThe LLM-based Kana-Kanji conversion systems have been an actively developing topic in the OSS community in recent years. That representative example is azooKey. It achieves high-quality conversion accelerated by an LLM. On the other hand, it has a clearly weak point.\r\nThis weak point is 'How to handle new words from context'. The existing method doesn’t do not use dictionaries tied to the LLM. They use a tokenizer before LLM inference, and their behavior is entirely dependent on the training data. One of the example of new words is \"Reiwa\". \"Reiwa\" is an era name in Japan. It was created in 2018.\r\nThe tokenizer splits dictionary words into multiple tokens. It loses the original meaning and part-of-speech (POS) information. As a result, these cases negatively impact the inference process for Kana-Kanji conversion.\r\n\r\nTo solve this issue, I propose an LLM extension method. This method bypasses the tokenizer and modified the LLM to accept inputs while preserving both the word's form and its POS information. After this modification, we can input feeding word data with attached POS tags directly into the LLM. When we use the new words in input, the system can derive conversion results that accurately reflect the intended usage. Yuino is the world first implementation of my proposed method.\r\nAttendees will learn about my proposed method deeply. The implementation of Yuino is a Japanese-specific implementation. On the other hand, the method is possible to reuse other languages in an ideogram user. Because my proposed method is common sense.",{"title":6028,"describe":6029,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FTK3BKR",{"id":6034,"room":6035,"start":3991,"end":3824,"language":96,"track":6036,"speakers":6038,"zh":6046,"en":6049,"tags":6050,"uri":6051},"TKXBBS",{"en":350,"zh-hant":10},{"id":3490,"name":6037},{"en":3492,"zh-hant":3493},[6039],{"id":6040,"avatar":6041,"zh":6042,"en":6045},"HYQJNJ","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FHYQJNJ_wWcNfkN.jpg",{"name":6043,"bio":6044},"農業開放資料社群COODFA","講者群：阿柴, 宜倩, Janice",{"name":6043,"bio":6044},{"title":6047,"describe":6048,"type":30},"我在農業現場: 雲林三小市集如何成為農業經理人","三小市集是「小人物、小土地、小滿足」的意思，我們從雲林出發，持續建立值得消費者信賴的品牌，成立十二年來，以市集、實體店面、電商、自媒體等之形式拓展客群，並開發自有品牌商品、發行雲林食通信雜誌、推出企業CSR好田計畫等，成為一個在地食農交流平台，讓生產與消費關係更正向地互動，目標讓生產者專心耕種、轉型友善栽培方式並獲得合理收入，讓在地農業與土地得以永續經營。",{"title":6047,"describe":6048,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FTKXBBS",{"id":6053,"room":6054,"start":4830,"end":4831,"language":96,"track":6055,"speakers":6057,"zh":6065,"en":6068,"tags":6071,"uri":6072},"TMBVKB",{"en":192,"zh-hant":10},{"id":196,"name":6056},{"en":198,"zh-hant":198},[6058],{"id":6059,"avatar":6060,"zh":6061,"en":6064},"XF8WBC","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FXF8WBC_XDoQ9vo.webp",{"name":6062,"bio":6063},"沈文傑","沈文傑，LottieFiles Senior Software Engineer，ThorVG 主要貢獻者之一。主要關注 SVG、Lottie、向量圖形渲染、C++ library API 設計，以及開源圖形函式庫在真實產品中的整合與維護。過去參與 ThorVG 的功能開發、相容性修正與 release 相關工作，希望透過這次分享，讓更多開發者理解設計素材進入應用程式後會遇到的工程問題。",{"name":6062,"bio":6063},{"title":6066,"describe":6067,"type":30},"ThorVG 1.0：SVG 與 Lottie 的跨平台渲染實務","SVG 和 Lottie 已經是常見的設計與動畫格式，但真正放進應用程式時，許多工程細節才會浮現：不同平台上的視覺結果是否一致、複雜動畫是否影響 UI 效能、設計素材如何被載入與重複使用、API 如何讓開發者修改場景中的元素。ThorVG 是 MIT 授權的開源 C++ 向量圖形引擎，支援 SVG、Lottie 與 2D rendering，v1.0 是專案邁向成熟應用的重要里程碑。\r\n\r\n這場分享會從 ThorVG 主要貢獻者的視角，介紹 SVG 與 Lottie 跨平台渲染背後的實作經驗。我會透過幾個實際案例說明：圖形與動畫資料如何進入繪製流程、相容性問題如何影響最後呈現、API 設計如何服務應用程式與 contributor，以及效能、檔案大小、穩定性與維護成本之間如何取捨。聽眾不需要圖形學背景，可以從 ThorVG 的例子理解一套開源基礎函式庫如何從「功能可用」逐步走向「能被真實產品依賴」。",{"title":6069,"describe":6070,"type":30},"ThorVG 1.0: Making SVG and Lottie Render Reliably Across Platforms","SVG and Lottie are common design and animation formats, but once they are used inside real applications, many details start to matter: whether graphics look consistent across platforms, whether complex animations slow down the UI, how assets are loaded and reused, and how APIs allow developers to modify elements inside the scene. ThorVG is an MIT-licensed open source C++ vector graphics engine for SVG, Lottie, and 2D rendering. Version 1.0 is an important milestone toward more mature real-world use.\r\n\r\nThis talk shares the engineering experience behind ThorVG 1.0 from the perspective of a core contributor. Through concrete examples, I will explain how SVG\u002FLottie parsing and compatibility issues affect the final image, how graphics data flows into rendering, how API design serves both application developers and contributors, and how performance, file size, stability, and maintenance cost are balanced. No graphics background is required. Attendees will see how an open source foundational library grows from “usable” to “reliable enough for real products\".",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FTMBVKB",{"id":6074,"room":6075,"start":4735,"end":3567,"language":70,"track":6076,"speakers":6078,"zh":6093,"en":6096,"tags":6097,"uri":6098},"TZCLYW",{"en":121,"zh-hant":10},{"id":4034,"name":6077},{"en":4036,"zh-hant":4036},[6079,6086],{"id":6080,"avatar":6081,"zh":6082,"en":6085},"7U9DWE","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002F7U9DWE_s666Wo0.webp",{"name":6083,"bio":6084},"Aung Kyaw Phyo","Aung Kyaw Phyo is the Founder and CTO of Binary Lab. With a strong passion for innovation and a track record of turning ideas into real products, he has played a key role in shaping the direction of Binary Lab and helping businesses and organisations leverage modern technology solutions. He leads a team of developers and technologists who share his drive for pushing boundaries in software and product development.\r\n\r\nAlongside his work at Binary Lab, he is currently collaborating with the Newsmast Foundation on Fediverse related projects, contributing to the growth of the open and decentralised social web. He is also actively exploring the space of AI and agentic systems, looking at how autonomous agents can be applied to real world software development and product workflows.",{"name":6083,"bio":6084},{"id":6087,"avatar":6088,"zh":6089,"en":6092},"NNFJL7","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FNNFJL7_HUTCXsw.webp",{"name":6090,"bio":6091},"Ye Myat Thu","React Native Team Lead",{"name":6090,"bio":6091},{"title":6094,"describe":6095,"type":30},"From One Codebase to Many Clients: How We Turn Mastodon Instances into No-Code Mobile Apps and help grow the Fediverse","## Overview\r\n\r\nThis session shares our technical journey of building a white-label Mastodon based mobile platform without forking Mastodon and without manually rebuilding apps for each client. We cover both the backend customisation layer and the mobile deployment pipeline, including the tradeoffs and lessons learned along the way. The talk is for developers working with ActivityPub, Mastodon, or mobile platforms who face the challenge to maintain custom features while keeping up with upstream changes.\r\n\r\n## Backend Challenges\r\n\r\nForking Mastodon and modifying the code directly works at first but becomes very painful as new versions come out. Keeping a fork up to date while preserving custom changes is time consuming, error prone, and does not scale across multiple clients.\r\n\r\n### Solution\r\n\r\nInstead of forking, we inject customisations through a purpose built Ruby Gem, keeping all modified code completely separate from the Mastodon repository. When a new version comes out, we update the gem instead of re-merging a fork.\r\n\r\n- Host a dedicated Mastodon instance for each client\r\n- Add or modify features through the Ruby Gem without touching core Mastodon code\r\n- Maintain separate gem versions for each Mastodon release\r\n\r\nWe also built a web based Dashboard so instance operators can control features without any code changes. Features shipped this way include Bluesky bridging, local only posts, long form post support, custom community feeds, and moderation tools.\r\n\r\n## Mobile Challenges\r\n\r\nWe manage multiple clients who all need the same core functionality, a Mastodon social feed with CMS news integration, but with different branding and feature variations. Manually updating icons, colours, and splash screens for every client did not scale. We needed to fully separate the core application from client configuration.\r\n\r\n### Solution\r\n\r\nWe built our architecture around three parts.\r\n\r\n**Control Center:** A web based platform where clients upload icons, choose colours, provide instance and CMS URLs, and configure features with no developer involvement needed.\r\n\r\n**Core App:** A React Native application with a scan to configure model. Scanning a QR code fetches a JSON configuration file and immediately rebrands the app at runtime without recompilation.\r\n\r\n**Build Service:** For changes that cannot be handled at runtime such as bundle IDs, app names, and push notification certificates, an automation pipeline manages custom builds and store deployment while keeping the no-code experience intact for the client.\r\n\r\n## Results\r\n\r\nThis architecture powers the Newsmast Foundation Apps for Change programme at newsmastfoundation.org, enabling rapid rollout of native apps across news outlets, content creators, professional networks, and local communities, growing the Fediverse through new and diverse audiences.\r\n\r\n## Key Takeaways\r\n\r\n- How combining Ruby Gems with a Dashboard lets you customise and ship features across multiple Mastodon instances without the burden of forking\r\n- How runtime configuration through QR code and JSON removes most manual work in multi-client mobile development\r\n- How better technical architecture can directly support an ambitious growth programme expanding the Fediverse globally",{"title":6094,"describe":6095,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FTZCLYW",{"id":6100,"room":6101,"start":4866,"end":4867,"language":70,"track":6102,"speakers":6104,"zh":6112,"en":6115,"tags":6117,"uri":6118},"TZQJJR",{"en":93,"zh-hant":10},{"id":3467,"name":6103},{"en":3469,"zh-hant":3469},[6105],{"id":6106,"avatar":6107,"zh":6108,"en":6111},"R7BPCU","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FR7BPCU_bvSrY9f.webp",{"name":6109,"bio":6110},"JaylinYu","Jaylin Yu graduated from CUHK, and previously worked as a Data analyst at UN ESCAP. holds several papers and patents on edge computing & networking. He has devoted himself to IoT, SDV and edge computing for more than ten years as a geek and has rich experience in the collaborative development of software and hardware integration in production. Now he is the Solution VP and initiator of the NanoMQ project in EMQ, a long-standing contributor to the open-source community (OASIS, LF, Eclipse, ES).",{"name":6109,"bio":6110},{"title":6113,"describe":6114,"type":30},"Open Source in Action: Pioneering Open-Source SDV in Real World production.","For OSS to fulfill its potential in Automotive, the industry must learn from past mistakes and successes, then chart a more strategic course forward. The author brings experience of applying OSS to real-world vehicles from 3 OEM projects, including V2X IoT infotainment system and data-closed loop platform across vehicle and cloud.\r\nIn this talk, the author shares lessons learned from the painful deployment process of the open-source SDV projects. It starts from a practical experience of in-vehicle performance tuning, to process signals of the whole vehicle (> 10,000) while only consuming below 2K DMIPS. It also unveils the methodologies for trigger-to-upload and transmission prioritization, ensuring guaranteed safety-related data delivery in remote diagnostic services, even under poor network conditions. This also covers topics of CI\u002FCD of OSS in mission-critical scenarios, by showcasing a story of building an in-vehicle real-time tracing feature for log and network packets. It serves as a workflow tool for better collaboration with other vendors, to escape from the blame game.\r\nThe session also gives an example of leveraging fuzzing technology to actively prevent intrusion.",{"title":6116,"describe":6114,"type":30},"開源軟體在行動：軟件定義汽車的實戰故事。",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FTZQJJR",{"id":6120,"room":6121,"start":3568,"end":3548,"language":96,"track":6122,"speakers":6124,"zh":6132,"en":6135,"tags":6138,"uri":6139},"TZTE7B",{"en":9,"zh-hant":10},{"id":3645,"name":6123},{"en":3647,"zh-hant":3648},[6125],{"id":6126,"avatar":6127,"zh":6128,"en":6131},"DYXRA7","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FDYXRA7_t4c9Ohi.webp",{"name":6129,"bio":6130},"ArBin","一個研究生",{"name":6129,"bio":6130},{"title":6133,"describe":6134,"type":30},"從 Page Reclaim 看 User-space Semantic Gap","在現代系統環境中，單一 Linux 主機常同時執行資料庫、VM、Container 與 Batch Job 等多種 Workload。然而，Kernel 在進行 Page Reclaim 時缺乏應用層語意，僅能依賴 Accessed bit、Refault 與 Page Type 等低階訊號推測資料冷熱，導致難以針對特定 Workload 做出最佳化決策。\r\n\r\n本議程將從 Linux reclaim path 與原始碼演進出發，探討 Active\u002FInactive LRU 與 Multi-Gen LRU（MGLRU）如何透過啟發式演算法推測 Page Hotness，以及這類設計在現代 Workload 下的限制。接著，將整理 madvise、posix_fadvise 與 userfaultfd 等機制，說明 Linux 如何讓 User-space 從單純提供 Hint，逐步走向參與記憶體決策路徑。\r\n\r\n最後將介紹近期 Linux 社群如何透過 eBPF 與 BPF-based memory policy patchset，嘗試將部分 Memory Management Decision 外部化，使記憶體管理從固定啟發式演算法，逐步走向可程式化決策框架，讓 Workload 語意能實際影響 Kernel 記憶體決策。",{"title":6136,"describe":6137,"type":30},"Examining the User-space Semantic Gap from Page Reclaim","In modern system environments, a single Linux host frequently executes diverse workloads concurrently, including databases, virtual machines (VMs), containers, and batch jobs. However, during Page Reclaim, the kernel lacks application-layer semantics and must rely exclusively on low-level signals—such as the Accessed bit, Refault, and Page Type—to infer data hotness. Consequently, achieving optimal memory management decisions for specific workloads remains challenging.\r\n\r\nThis session will begin by examining the Linux reclaim path and the evolution of its source code. It will explore how the Active\u002FInactive LRU and Multi-Gen LRU (MGLRU) utilize heuristic algorithms to infer page hotness, alongside the limitations inherent to these designs under modern workloads. Subsequently, the session will review mechanisms such as madvise, posix_fadvise, and userfaultfd, illustrating how Linux enables User-space to transition from merely providing hints to actively participating in the memory decision-making path.\r\n\r\nFinally, the session will introduce recent efforts by the Linux community to externalize specific memory management decisions using eBPF and BPF-based memory policy patchsets. This demonstrates a paradigm shift in memory management, moving from fixed heuristic algorithms toward a programmable decision framework, thereby allowing workload semantics to directly influence kernel memory decisions.",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FTZTE7B",{"id":6141,"room":6142,"start":3504,"end":3505,"language":70,"track":6143,"speakers":6145,"zh":6153,"en":6156,"tags":6157,"uri":6158},"UGAQP3",{"en":121,"zh-hant":10},{"id":4034,"name":6144},{"en":4036,"zh-hant":4036},[6146],{"id":6147,"avatar":6148,"zh":6149,"en":6152},"X93LWP","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FX93LWP_YnWIQCb.webp",{"name":6150,"bio":6151},"Roro Park","Roro Park is the founder of East Asia Technology Institute (EATI), a Seoul-based research and advisory organization focused on technology governance, geopolitics, and regional policy analysis across Korea, Japan, Taiwan, and the wider East Asian region.\r\n\r\nWith a background in film, psychology, and music production, Roro brings an interdisciplinary perspective to technology, culture, communication, and governance. Her current work examines how open and emerging technologies can be adapted responsibly across different legal, social, and regional contexts.",{"name":6150,"bio":6151},{"title":6154,"describe":6155,"type":30},"Federation Is Not Enough: Governance Patterns for the Open Social Web in East Asia","The Fediverse is often described as a technical alternative to centralized social platforms: decentralized, interoperable, portable, and community-governed. But for the open social web to grow meaningfully in East Asia, protocol compatibility alone is not enough. Federated systems also need governance patterns that can work across different legal environments, platform cultures, languages, and community expectations.\r\n\r\nThe East Asia Technology Institute proposes a talk that examines the Fediverse and open social web from an East Asian governance perspective, with a focus on Korea, Japan, and Taiwan. Rather than treating the region as one market, it looks at how different privacy regimes, moderation expectations, localisation needs, and trust cultures can shape the way federated social systems are built and operated.\r\n\r\nThe session is not intended as legal advice. Instead, it translates governance questions into practical considerations for developers, instance operators, moderators, open-source contributors, and community oraganisers. Topics may include instance accountability, data-flow transparency, cross-border federation, moderation processes, user onboarding, localisation, and the role of open standards in avoiding vendor lock-in.\r\n\r\nThe central argument is that the Fediverse should not treat governance as an afterthought. If open social infrastructure is to become genuinely global, it must be able to adapt to regional realities while preserving interoperability, openness, and community control.",{"title":6154,"describe":6155,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FUGAQP3",{"id":6160,"room":6161,"start":3464,"end":3465,"language":1469,"track":6162,"speakers":6164,"zh":6172,"en":6175,"tags":6176,"uri":6177},"UKZ8LS",{"en":121,"zh-hant":10},{"id":4034,"name":6163},{"en":4036,"zh-hant":4036},[6165],{"id":6166,"avatar":6167,"zh":6168,"en":6171},"S9ETNQ","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FS9ETNQ_1ChY35n.webp",{"name":6169,"bio":6170},"Itoh Shimon","https:\u002F\u002Fpostfixnotation.org\u002F",{"name":6169,"bio":6170},{"title":6173,"describe":6174,"type":30},"Vertical writing for the Mongolian script on Mastodon","The Mongolian script is mainly used for Mongolian. The script is in daily use in the Inner Mongolia Autonomous Region of the People's Republic of China. The script mainly used in Mongolia is the Cyrillic script, but the Mongolian script is being reintroduced in Mongolia.\r\n\r\nThe Mongolian script should be displayed vertically, but vertical writing for the script is not well supported in most software. In particular, most global social media do not support vertical writing for the script. Therefore, Mongolian script users write sideways and post images containing Mongolian script text.\r\n\r\nTherefore, I proposed and implemented partial support for the Mongolian script on Mastodon. I made it possible to display Mongolian script text vertically without changing the UI layout. This approach attracted attention and influenced other fediverse software. New users joined Mastodon because Mastodon supports vertical writing for the Mongolian script.\r\n\r\nThis talk is about the Mongolian script, issues that occurred in this approach, and the influence of this approach. This approach for internationalization demonstrates advantages of the Fediverse over proprietary global social media.",{"title":6173,"describe":6174,"type":30},[],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FUKZ8LS",{"id":6179,"room":6180,"start":4536,"end":4054,"language":96,"track":6181,"speakers":6183,"zh":6191,"en":6194,"tags":6197,"uri":6198},"USZCJN",{"en":192,"zh-hant":10},{"id":196,"name":6182},{"en":198,"zh-hant":198},[6184],{"id":6185,"avatar":6186,"zh":6187,"en":6190},"M7C8Q9","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FM7C8Q9_kygNTl7.webp",{"name":6188,"bio":6189},"許育瑋","Ave Mujica 演唱會抽到票了,真是奇蹟\r\n\r\n21 歲，是個學生\r\n\r\n修了 Linux 核心設計之後踏上了探索系統軟體的旅途。\r\n之前把系上必修課做的奇怪「微電腦」搬到 SITCON Demo 展出,最近在研究 SCX 和複習統計學。\r\n\r\nGitHub: [weiso131](https:\u002F\u002Fgithub.com\u002Fweiso131)",{"name":6188,"bio":6189},{"title":6192,"describe":6193,"type":30},"藉由 sched_ext 客製化低延遲 CPU 排程器","「為什麼遊戲又突然卡了一下？」的答案往往並非 CPU 不夠快，而是 CPU 排程器把運算資源分給那些遊戲玩家根本不在意的背景任務。\r\n\r\nLinux 的 EEVDF 排程器以 fairness 為主要目標，並透過 EWMA 估計負載。然而，fairness 不等於流暢體感、EWMA 只能反映短期變化，因此延遲敏感的工作負載常出現不穩定的表現。`scx_teddy` 是一款針對遊戲場景、建構在機器學習之上的 CPU 排程器，利用 Linux v6.12+ 引入的 `sched_ext` (SCX)，其目標不是要求使用者手動指定重要任務，而是讓系統自行辨識哪些是影響遊戲流暢度的關鍵操作、哪些只是可退讓的背景任務，進而依據遊戲期望調整。\r\n\r\n本系統會在 Linux 核心內部長期觀測任務行為特徵，如 runtime、sleep、iowait 等，再於使用者層級藉由 K-means 進行非監督式分群。完成分類後，`scx_teddy` 便可依據針對遊戲情境設計的 cluster policy，動態調整優先級、time slice，以及是否獨佔處理器。\r\n\r\n本議程將介紹 `scx_teddy` 的設計與實作，包括：\r\n* 為何選擇 ML，而非規則式分類\r\n* 特徵工程與行為指標的取捨\r\n* eBPF 與 user space 的資料管線設計\r\n* 與既有 SCX 排程器（如 `scx_cake`）的差異與實驗結果\r\n\r\n目標不是追求更高的 benchmark 分數，而是讓系統真正理解：哪些工作會影響玩家體感，哪些則不值得搶佔遊戲的執行時間。",{"title":6195,"describe":6196,"type":30},"Customizing a Low-Latency CPU Scheduler with sched_ext","\"Why did my game just stutter again?\" The answer is often not insufficient CPU power, but a scheduler distributing compute resources to background tasks the gamer doesn't care about.\r\nLinux's EEVDF scheduler optimizes primarily for fairness and estimates load via EWMA. However, fairness does not equal a smooth user experience, and EWMA only captures short-term fluctuations, leaving latency-sensitive workloads with unstable behavior. scx_teddy is a gaming-focused, machine-learning-based CPU scheduler built on sched_ext (SCX), introduced in Linux v6.12+. Rather than requiring users to manually mark important tasks, it lets the system itself identify which threads are critical to gameplay smoothness and which are merely background tasks that can step aside, then adjusts scheduling decisions according to the gamer's expectations.\r\nThe system observes per-task behavioral features (runtime, sleep ratio, iowait, etc.) in the Linux kernel over long time windows, then performs unsupervised K-means clustering in user space. Once tasks are classified, scx_teddy applies a cluster policy designed for gaming scenarios, dynamically adjusting priority, time slice, and whether a cluster gets exclusive access to performance cores.\r\nThis talk presents the design and implementation of scx_teddy, including:\r\n\r\n* Why ML over rule-based classification\r\n* Trade-offs in feature engineering and behavioral metrics\r\n* Data pipeline design between eBPF and user space\r\n* Comparison and experimental results against existing SCX schedulers such as scx_cake\r\n\r\nThe goal is not to chase higher benchmark scores, but to let the system genuinely understand which tasks affect what the player feels — and which don't deserve to preempt the game.",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FUSZCJN",{"id":6200,"room":6201,"start":3505,"end":4536,"language":328,"track":6202,"speakers":6204,"zh":6212,"en":6215,"tags":6216,"uri":6217},"UY8UQN",{"en":263,"zh-hant":10},{"id":44,"name":6203},{"en":46,"zh-hant":47},[6205],{"id":6206,"avatar":6207,"zh":6208,"en":6211},"X3GX3V","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FX3GX3V_Gfhz9LO.webp",{"name":6209,"bio":6210},"Raghu","I'm primarily a Software Dev on the backend, but also dabble around in some microcontroller programming (to troll iOS users) and security research. I love networking - this means IP, TCP, Wireshark etc. , not talking business at cocktail parties (though I like cocktails otherwise).\r\n\r\nWhen I'm not behind my keyboard I like to travel to and cross weird borders - my favorite so far has been a hovercraft across the frozen Heilongjiang river from Heihe to Blagoveshchensk.\r\n\r\nI use arch, btw",{"name":6209,"bio":6210},{"title":6213,"describe":6214,"type":59},"The Workings of the Internet or: How Censors Block You From Websites","\"The Internet is a big network of computers all over the world connected to each other\" - most of us have probably come across an explanation such as this when trying to learn how it all works. However, this does not explain anything at a technical level – what is really happening when you type “google.com” in your browser and hit enter?\r\n\r\nWhy does “google.com” work, but some, ahem, other websites get blocked? How? Can the person sitting next to you in a coffee shop really steal all your data unless you use “CompanyVPN”? (Don’t forget to use code TOR_OR_BUST for 1337% off!)\r\n\r\nThere is a lot that goes into making the internet work, and there are a lot of (mostly legitimate) parties between you and the website you’re visiting. Understanding how it all works is key to understanding how this system is (ab)used to perform surveillance, block access, and steal data. \r\n\r\nThis talk aims to first explain some basic concepts of how two parties can communicate over the internet via “packets”, analogous to postcards (ever sent those?!). Through this analogy, we’ll learn exactly who sits in the middle, what can they see and can they change.  You'll come to appreciate how brittle parts of Internet infrastructure can be, the number of parties that need to cooperate across the world, and maybe end up surprised that any of it works at all.\r\n\r\nWe’ll also explore how bad actors can tamper with the “Postcards” - explaining how censors work, or how information is leaked, even in 2026 with all the decades of improvements since the 80s.\r\nBy the end, you’ll either be surprised to learn how private the internet is by default, or end up annoyed at how some stuff still gets out! And of course, you will start judging the predatory VPN advertising more harshly.",{"title":6213,"describe":6214,"type":59},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FUY8UQN",{"id":6219,"room":6220,"start":3754,"end":3755,"language":96,"track":6221,"speakers":6223,"zh":6250,"en":6253,"tags":6256,"uri":6257},"VA3UPR",{"en":1253,"zh-hant":10},{"id":901,"name":6222},{"en":903,"zh-hant":904},[6224,6231,6238,6245],{"id":6225,"avatar":6226,"zh":6227,"en":6230},"FQWPA9","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FFQWPA9_uQoDujZ.webp",{"name":6228,"bio":6229},"Yider Hsu","講者覺得放MLC AIRR MM Taiwan Team比較好 \r\nMLCommons 是一個成立於 2020 年的開放式 AI 工程聯盟，是橫跨產、官、學界的成員組成，其核心使命是透過公開透明的基準測試、數據集與實踐規範，加速人工智慧與機器學習的創新並普及惠及全球。 其發起的 AILuminate 文化特定多模態基準測試（AILuminate Culturally-Specific Multimodal Benchmark） 計畫透過與各地的學術、工業及政府機構與組織合作，形成 AI 風險與可靠性（AI Risk & Reliability, AIRR）工作小組AIRR，工作小組內又細分多個小組，我們是關注文字、影像等多形式數據的安全評估的，台灣多模態(Multimodal)小組。\r\nTaiwan Team 大部分的成員是coscup的義工，想將現階段的工作分享給大家。",{"name":6228,"bio":6229},{"id":6232,"avatar":6233,"zh":6234,"en":6237},"KFBXDN","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FKFBXDN_pE3X5wx.webp",{"name":6235,"bio":6236},"Tom Lin","目前從事與地理相關的工作。\r\n\r\n畢業於地理學系，研究興趣為歷史地理學、區域地理及臺灣史。大學因修課的關係，而萌生出關注議題有文化研究及性別議題等人文社會領域範疇，並因為參與自然保育相關的學術性社團，也有關心自然環境相關等議題。\r\n\r\n從大學到研究所跨足不同領域——從「人類活動」到「地理空間」，但核心概念仍是以「人」為出發點。「文化」是由人類創造的，兼具有形和無形；「地理學」的學科傳統之一為「人地關係」，即人與環境互動的過程，也是離不開人。\r\n\r\n吾身為生活在這塊土地上的「人」，無論對於「人」，甚至是「土地」的種種，除了要提醒自己保持著好奇的求知慾，以及柔軟和開闊的心胸面對物理性和化學性的生活變化，這也是一輩子的人生功課。",{"name":6235,"bio":6236},{"id":6239,"avatar":6240,"zh":6241,"en":6244},"WHGDAG","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FWHGDAG_ZQP1tjR.jpg",{"name":6242,"bio":6243},"Jeff Huang","* 中華民國軟體自由協會理事\r\n* 維基媒體運動貢獻者\r\n* 自由軟體譯者\r\n* 文件基金會會員\r\n* KDE e.V. 會員\r\n* 開放街圖基金會會員",{"name":6242,"bio":6243},{"id":6246,"avatar":3,"zh":6247,"en":6249},"UHYCMT",{"name":6248,"bio":10},"petertc",{"name":6248,"bio":10},{"title":6251,"describe":6252,"type":30},"AI模型的文化盲區─探討AI測試標準的開源現況觀察與難題","AI模型的文化盲區─探討AI測試標準的開源現況觀察與難題\r\n\r\n目前的 AI 模型雖具備高度的技術流暢性， 但長期局限於西方觀點的現狀，缺乏對全球語言多樣性的理解，導致模型在面對非西方語境時容易產生判斷失誤，看似無害的文字與影像結合卻可能觸犯特定文化禁忌，為了彌補這一文化盲點，一個名為 AI Luminate 的計畫產生。由開放式 AI 工程聯盟MLCommons 主導，涵蓋印度、新加坡、韓國、日本、孟加拉、台灣等不同團隊，透過團隊開發「多模態安全測試套件」AIRR Multimodal ，來提升模型對多語系環境中社交與商業細微差異的應變能力。透過這種跨國界的合作評估來確保 AI 能在多元的現實世界中，具備文化智力與全球適用性的安全基準，成為對數十億使用者均安全且有效的知識介面。\r\n\r\n但實際上人工智慧在處理多元文化語境與安全性評估時，往往在理解非英語使用地區的社交禁忌與文化細微差別上表現不佳，在面對非西方背景的情境容易產生判斷失誤，存在極高的漏報率。要如何使AI 的能真正理解多元文化語系背後的文化意涵與潛在衝突，打造具備文化智商的全球化 AI，是目前所面臨的嚴峻挑戰。",{"title":6254,"describe":6255,"type":30},"The Cultural Blind Spot of AI Models: Observing the Open-Source Status and Challenges of AI Testing Standards","While current AI models possess a high degree of technical fluency, they are often limited by Anglosphere perspective and lack an understanding of global linguistic diversity. This results in models easily making judgment errors when facing non-Western contexts. What appears to be a harmless combination of text and image might, in fact, violate specific cultural taboos. To address this cultural blind spot, a project named AI Luminate was initiated. Led by the open AI engineering consortium MLCommons, and involving different teams from India, Singapore, South Korea, Japan, Bangladesh, and Taiwan, the project is developing the \"Multimodal Safety Test Suite,\" AIRR Multimodal, to enhance the models' ability to respond to social and commercial nuances in multilingual environments. This cross-border cooperative assessment aims to ensure AI has cultural intelligence and a globally applicable safety standard in a diverse real world, becoming a safe and effective knowledge interface for billions of users.1\r\n\r\n\r\nIn reality, however, when handling diverse cultural contexts and safety assessments, artificial intelligence often performs poorly in understanding social taboos and cultural subtleties in non-English-speaking regions. This results in models easily making judgment errors when facing non-Western background scenarios, leading to an extremely high false negative rate. The severe challenge currently faced is how to enable AI to truly understand the cultural implications and potential conflicts behind diverse linguistic and cultural systems, thereby creating a globalized AI with cultural intelligence.",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FVA3UPR",{"id":6259,"room":6260,"start":3709,"end":3823,"language":13,"track":6261,"speakers":6263,"zh":6271,"en":6274,"tags":6275,"uri":6276},"VBFGUL",{"en":1063,"zh-hant":10},{"id":1065,"name":6262},{"en":1067,"zh-hant":1068},[6264],{"id":6265,"avatar":6266,"zh":6267,"en":6270},"N8XMH8","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002Fmaeda_0wz1VXu.jpg",{"name":6268,"bio":6269},"MAEDA Go","MAEDA Go (前田剛) is a Redmine committer and long-time contributor from Japan.\r\n\r\nHe has been using Redmine since 2007 and founded Redmine.JP, one of the largest Japanese information websites for Redmine. He has also contributed to the Redmine community through books and technical articles about Redmine.\r\n\r\nHe became a Redmine committer in 2017 and continues to contribute to the project and its community.",{"name":6268,"bio":6269},{"title":6272,"describe":6273,"type":30},"Adding Right-to-Left support to your web application with CSS logical properties — Lessons from Redmine","Supporting right-to-left (RTL) languages such as Arabic and Hebrew remains an often overlooked challenge in web application internationalization. Unlike English and Chinese, which are written left to right (LTR), RTL languages require layouts and visual alignment throughout the user interface to be mirrored.\r\n\r\nRedmine, a 20-year-old open source project management application, has supported RTL languages since 2010. Its RTL support was originally implemented using a separate RTL-specific stylesheet that overrode the main LTR styles, which was a common approach at the time. As the user interface evolved, maintaining two parallel stylesheets became increasingly difficult, and the RTL layout gradually became broken in many places.\r\n\r\nIn this talk, I will explain how I rebuilt Redmine's RTL support using CSS logical properties. By replacing physical properties such as `margin-left` and `padding-right` with logical properties, a single stylesheet can automatically adapt to both LTR and RTL layouts. I will discuss the migration process, practical implementation techniques, and lessons learned while modernizing an existing codebase.\r\n\r\nAttendees will learn how existing LTR-only web applications can gain RTL support with a relatively small set of CSS changes. Through practical examples from Redmine, they will learn concrete techniques for implementing maintainable RTL support using CSS logical properties. Although the examples come from Redmine, the approach is applicable to many web applications looking to add maintainable RTL support.",{"title":6272,"describe":6273,"type":30},[],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FVBFGUL",{"id":6278,"room":6279,"start":3909,"end":3910,"language":96,"track":6280,"speakers":6282,"zh":6290,"en":6293,"tags":6296,"uri":6297},"VE8MGP",{"en":192,"zh-hant":10},{"id":196,"name":6281},{"en":198,"zh-hant":198},[6283],{"id":6284,"avatar":6285,"zh":6286,"en":6289},"7L7AG8","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002F7L7AG8_nMqA8gF.webp",{"name":6287,"bio":6288},"蕭于恩","畢業於國立高雄大學資訊系（2022 年），致力於投入系統軟體開發，目前在編譯器領域累積了一點點成果",{"name":6287,"bio":6288},{"title":6291,"describe":6292,"type":30},"動態連結的考量和實踐：以 shecc 為例","[shecc](https:\u002F\u002Fgithub.com\u002Fsysprog21\u002Fshecc) 是個精簡的 C 語言最佳化編譯器，以約一萬行 C 程式實作出完整編譯流程，涵蓋語法解析、中間表示 (IR)、SSA (static single assignment form)、經典最佳化，及程式碼產生。在不依賴外部工具的前提下，可直接產生 Arm 與 RISC-V 架構的執行檔，並具備自我編譯 (self-hosting) 能力。\r\n\r\n最初 shecc 僅支援靜態連結輸出，自 2025 年起，逐步導入動態連結機制，使產生的執行檔可於執行時期連結 glibc，進而使用更完整的 C 標準函式庫。本議程將以 shecc 為例，說明如何將僅支援靜態連結的編譯器演進為支援動態連結的系統，並探討以下關鍵議題：\r\n\r\n* 為 ELF 執行檔產生 dynamic sections，使其成為可動態連結的格式\r\n* 解析函式呼叫於執行時期如何透過 PLT\u002FGOT 與動態連結器完成重定位，並正確導向 glibc 實作\r\n* 改進 Application Binary Interface (ABI)，尤其針對 Arm 架構，使函式呼叫符合規範並正確運作",{"title":6294,"describe":6295,"type":30},"Dynamic Linking Considerations and Implementations: A Case Study of shecc","[shecc](https:\u002F\u002Fgithub.com\u002Fsysprog21\u002Fshecc) is a self-hosting, optimizing C compiler that integrates a parser, Intermediate Representation (IR), Static Single Assignment Form (SSA), optimization algorithms, and a code generator within approximately 14000 lines of C code. It is capable of directly generating ELF executables for Arm or RISC-V architectures without relying on external assemblers or linkers.\r\n\r\nInitially, shecc only supported the generation of statically linked executables. However, as of 2025, shecc has gained the capability to produce dynamically linked programs, allowing them to work with a fully functional C standard library (glibc) at runtime. This presentation illustrates the enhancements made to support dynamic linking and discuss the following key issues:\r\n\r\n* Integrating dynamic sections into ELF executables.\r\n* Implementing PLT\u002FGOT mechanisms to facilitate runtime relocation via the dynamic linking.\r\n* Improving Application Binary Interface (ABI) compatibility, specifically for the Arm architecture, to ensure the correctness and reliability of function calls.",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FVE8MGP",{"id":6299,"room":6300,"start":3969,"end":4447,"language":70,"track":6301,"speakers":6303,"zh":6311,"en":6314,"tags":6315,"uri":6316},"VGLTAB",{"en":1253,"zh-hant":10},{"id":901,"name":6302},{"en":903,"zh-hant":904},[6304],{"id":6305,"avatar":6306,"zh":6307,"en":6310},"UKGMA8","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FUKGMA8_LssM07R.webp",{"name":6308,"bio":6309},"Shirley","Shirley is a software engineer at Grafana Labs working on solutions to facilitate observability. She has lately been curious about learning to solder and making blinky LED badges. Outside of work, she enjoys bike riding, baking, and knitting.",{"name":6308,"bio":6309},{"title":6312,"describe":6313,"type":30},"My Flipper Zero is Useless (and Other Silly Stories)","The Flipper Zero has a reputation as a hacker’s multitool capable of cloning keycards, starting Teslas, and stealing WiFi passwords. Curious about the hype, I bought one to see what it could actually do.\r\n\r\nIn this talk, I’ll introduce the Flipper Zero and explore the wireless technologies it interacts with, including infrared, RFID, and NFC. Along the way, we’ll explore how access cards work and why many real-world systems are harder to clone than internet hype suggests.\r\n\r\nBut the most surprising part of my Flipper Zero journey wasn’t hacking, it was discovering the joy of building software for a tiny open device. I’ll share a few small projects, including my first simple knit counter app and a silly scorekeeping app using the Flipper Zero, Grafana dashboard, Google Sheets as a stack.\r\n\r\nThis talk is about curiosity, creativity, and my journey as a so-called hacker who set out to hack things, failed completely, and ended up building wholesome and unexpected projects instead.",{"title":6312,"describe":6313,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FVGLTAB",{"id":6318,"room":6319,"start":3754,"end":3755,"language":96,"track":6320,"speakers":6322,"zh":6330,"en":6333,"tags":6336,"uri":6337},"VJXJZ8",{"en":282,"zh-hant":10},{"id":3670,"name":6321},{"en":3672,"zh-hant":3673},[6323],{"id":6324,"avatar":6325,"zh":6326,"en":6329},"38MTQ9","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002Fjimmy-eye_2_qCb123u.jpeg",{"name":6327,"bio":6328},"Jimmy Liao","https:\u002F\u002Fsessionize.com\u002Fjimmy-liao\u002F",{"name":6327,"bio":6328},{"title":6331,"describe":6332,"type":30},"Multi-Agent Digest 實戰：用 ADK × Gemini Enterprise Agent Platform 把 4  個 Agent 推上企業級平台","這幾年大家都在談 Multi-Agent，但從「LangGraph 寫範例」到「真的部署在雲端、用戶端跑得起來」之間，差了一整個 production gap。\r\n\r\n 這場 talk 會以開源專案 jimmyliao\u002Fdigest-agent (MIT) 為主軸，分享一個從 Streamlit MVP 升級成 Next.js + ADK Multi-Agent 的完整改造旅程，重點放在架構決策：\r\n\r\n1. **為什麼選 Google ADK 而不是自寫 orchestration**\r\n2. **Gemini Enterprise Agent Platform (GEAP) 跟 Cloud Run 的職責切分**\r\n\r\n現場會 demo 一個跑在 Gemini Enterprise Agent Engine 上的即時 multi-agent 分析串流。",{"title":6334,"describe":6335,"type":30},"Multi-Agent Digest with ADK × GEAP (Gemini Enterprise Agent Platform)","Going from a LangGraph notebook to a production-grade\r\n  Multi-Agent system on cloud is a huge gap. This talk\r\n  walks through the open-source jimmyliao\u002Fdigest-agent\r\n  project — a Streamlit MVP rebuilt as Next.js + ADK\r\n  SequentialAgent (4 sub-agents) deployed on Vertex\r\n  Agent Engine (GEAP), with Cloud Run UI streaming via\r\n  SSE. Three architectural decisions: (1) why Google\r\n  ADK over hand-rolled orchestration, (2) splitting\r\n  responsibility between Cloud Run and GEAP Agent\r\n  Runtime, (3) using SQLite + Litestream for restart-\r\n  safe state. Live demo: a 4-agent stream analyzing\r\n  TSMC vs MediaTek vs Foxconn. All code MIT-licensed,\r\n  runnable from a Cloud Shell `make` command. Validated\r\n  in a 90-min hands-on workshop at BwAI Taichung 2026.",[402],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FVJXJZ8",{"id":6339,"room":6340,"start":3709,"end":4054,"language":96,"track":6341,"speakers":6343,"zh":6347,"en":6350,"tags":6351,"uri":6352},"VLL8PX",{"en":684,"zh-hant":10},{"id":686,"name":6342},{"en":688,"zh-hant":689},[6344],{"id":2156,"avatar":2157,"zh":6345,"en":6346},{"name":2159,"bio":2160},{"name":2159,"bio":2160},{"title":6348,"describe":6349,"type":30},"揭秘審查演算法：談平台治理政策之前，先來開源平台的審查演算法吧","為何你的臉書帳號會無故被封鎖？為何會被降觸及？詐騙留言、詐騙貼文為何層出不窮，政府、平台難道不管管嗎？這些都是平台治理的核心問題。但是在這之前，使用者（或消費者）對於平台審查演算法可說是一無所知。另一方面，平台並不會主動揭露自己的排序或審查演算法，因為這將會讓惡意帳戶攻克平台築起來的防詐城牆。因此開源與封閉排序與審查演算法，可說是都有理由。\r\n\r\n本次討論將揭露華文抗審查文章平台 Matters 的審查演算法，我們同時將演算法開源，且進一步開放違規內容的資料庫，同時進行平台治理的社群實驗。目的在於讓有興趣參與台灣平台治理的開源貢獻者，能提早一步理解平台營運者怎麼思考這些問題，讓未來平台治理的倡議可以更加落地。",{"title":6348,"describe":6349,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FVLL8PX",{"id":6354,"room":6355,"start":3754,"end":3755,"language":328,"track":6356,"speakers":6358,"zh":6366,"en":6369,"tags":6370,"uri":6371},"VQNU3N",{"en":522,"zh-hant":10},{"id":3593,"name":6357},{"en":3595,"zh-hant":10},[6359],{"id":6360,"avatar":6361,"zh":6362,"en":6365},"RYFKX3","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FRYFKX3_GlLYOxl.webp",{"name":6363,"bio":6364},"Nian-Ze Lee","Dr. Nian-Ze Lee is an Assistant Professor in the Department of Electrical Engineering at National Taiwan University (NTUEE), a position he has held since February 2025, and a Visiting Professor at the Institute of Informatics, LMU Munich, Germany. He earned his Ph.D. in Electronics Engineering from NTU in 2021, where he was selected as an honorary member of the Phi Tau Phi Scholastic Honor Society. His dissertation on stochastic Boolean satisfiability received the prestigious Lam Research Thesis Award. From 2021 to 2024, Dr. Lee was a Postdoctoral Researcher at LMU Munich, Germany, focusing on formal methods and their practical applications in software engineering and EDA. His work has been recognized at top-tier international conferences and earned several awards, including an ETAPS Distinguished Paper Award at TACAS 2026, an ACM SIGSOFT Distinguished Paper Award and a Best Artifact Award at FSE 2024, a Best Paper Award at SPIN 2024, and a Distinguished Artifact Award at TACAS 2024. As of 2026, he leads multiple research projects as a principal investigator, including a DFG-funded research project and an Intel-funded collaboration. From 2025 to 2028, he holds the Garmin Scholar Fellowship at NTU.",{"name":6363,"bio":6364},{"title":6367,"describe":6368,"type":30},"A Case Study in Firmware Verification: Applying Formal Methods to Intel® TDX Module","Firmware underpins system security but remains challenging to verify due to hardware dependency, specialized coding idioms, and limited open-source examples. Manual verification approaches, while common in industry, are labor-intensive and difficult to scale. This paper presents a detailed case study on applying automatic formal methods for software to a security-critical firmware component in Intel Trust Domain Extensions (TDX), known as TDX Module. In this study, we employ six state-of-the-art C-program analyzers on the production TDX Module firmware, leveraging techniques ranging from bounded model checking and symbolic execution to abstract interpretation. Our empirical evaluation identifies obstacles unique to firmware, highlights harness-design decisions essential for verifying industry-scale code bases, and demonstrates opportunities in advanced slicing for more scalable verification. Although the case study focuses on TDX Module, the findings are broadly applicable to large-scale, low-level programs and have already influenced the software-verification community, such as standardizing nondeterministic object initialization. All verification tasks and proof harnesses are publicly released to foster reproducible research and future tool development.",{"title":6367,"describe":6368,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FVQNU3N",{"id":6373,"room":6374,"start":3910,"end":3968,"language":96,"track":6375,"speakers":6377,"zh":6385,"en":6388,"tags":6391,"uri":6392},"VTFUFD",{"en":922,"zh-hant":10},{"id":330,"name":6376},{"en":332,"zh-hant":332},[6378],{"id":6379,"avatar":6380,"zh":6381,"en":6384},"9FARZ9","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002F9FARZ9_pgJQjTY.webp",{"name":6382,"bio":6383},"李世元 \u002F Shih-Yuan Lee \u002F FourDollars \u002F $4","用 Linux 超過 26 年、從事 Linux 相關工作 21 年，其中 16 年在 Canonical。主要在做「讓 Ubuntu 能順順在各種筆電上跑起來」這件事。日常工作包括跟硬體廠商打交道、追各種奇怪的驅動程式 bug，以及建一堆自動化讓大家不用手動做重複的事。閒暇之餘會貢獻開源專案，有時候是修 kernel，有時候是跟上游維護者吵架然後說服對方接受你的 patch。是 Debian 開發者，主要意義是對套件命名有很多意見。\r\n\r\n目前正在尋找新機會。如果你在找一個能把 Linux 跑好、跟硬體廠商打過交道、又會把重複的事自動化掉的人，歡迎聊聊。\r\n\r\nA Linux user for over 26 years and professional for 21, Shih-Yuan has spent the last 16 of those at Canonical making sure Ubuntu actually works on the laptops people buy. That means working with hardware vendors, chasing down driver bugs, and building the automation that keeps everything running quietly in the background. He contributes to open source projects when he can — sometimes patching the kernel, sometimes getting into a good argument with upstream maintainers. He's a Debian Developer, which mostly means he has opinions about package naming.\r\n\r\nHe's currently open to new opportunities. If you need someone who can make Linux behave on real hardware — and automate away the parts that don't — he'd be happy to talk.",{"name":6382,"bio":6383},{"title":6386,"describe":6387,"type":59},"動手玩 OpenClaw：在你的 Ubuntu 筆電上用 Juju + LXD 架出自己的 Telegram AI 機器人","這是一場帶著電腦動手做的 Workshop，目標是讓每位參與者在 50 分鐘內，在自己的 Ubuntu 機器上用 Juju + LXD 跑起 OpenClaw，並透過 Telegram 機器人收到第一個 AI 回覆。本 Workshop 以 Google AI Studio（ai.dev）提供的免費 API key 作為 AI 來源，參與者只需要一個 Google 帳號，無需任何付費訂閱即可完整體驗。課程分為四個段落：\r\n（1）環境準備——確認 Ubuntu 24.04、安裝 LXD snap 與 Juju，初始化 local cloud；\r\n（2）部署 OpenClaw Charm——juju add-model、juju deploy openclaw、填入從 Google AI Studio 取得的免費 API key；\r\n（3）整合 Telegram——在 BotFather 建立 bot 取得 token 後填入 juju config telegram-bot-token，接著傳訊息給 bot 並取得 AI 回覆；\r\n（4）進階調整與觀察——使用 openclaw tui 直接在終端機與 AI 助理互動，即時觀察對話狀態與回應。\r\n\r\n所有步驟皆有事先準備好的指令清單，講師與助教（如果有找到人的話）在現場協助排錯。先決條件：原生 Ubuntu 24.04 的筆電或實體\u002F虛擬機器（WSL2 不支援作為 LXD host，建議使用原生 Ubuntu 或 Ubuntu VM）、網路存取、Google 帳號、Telegram 帳號。\r\n\r\nMore info: https:\u002F\u002Fevents.canonical.com\u002Fevent\u002F146\u002Fcontributions\u002F931\u002F",{"title":6389,"describe":6390,"type":59},"Hands-On OpenClaw: Build Your Own Telegram AI Bot on Ubuntu with Juju and LXD","This is a bring-your-laptop workshop designed to help every participant get OpenClaw running on their own Ubuntu machine within 50 minutes using Juju and LXD—and receive their first AI reply over Telegram. This workshop uses a free API key from Google AI Studio (ai.dev) as the AI source, so all you need is a Google account—no paid subscription required. The session covers four segments: \r\n(1) Environment setup—Ubuntu 24.04, LXD snap, Juju installation, and local cloud initialization; \r\n(2) Deploying the OpenClaw Charm—running juju add-model, juju deploy openclaw, and setting the free API key obtained from Google AI Studio; \r\n(3) Telegram integration—creating a bot via BotFather, applying the token with juju config telegram-bot-token, and chatting with the bot live; \r\n(4) Advanced tweaks and observation—using openclaw tui to interact with the AI assistant directly from the terminal and observe the conversation in real time. \r\n\r\nAll steps come with a pre-prepared command cheat sheet, and the instructor (and helpers, if available) will assist on the floor. Prerequisites: A machine running native Ubuntu 24.04 (laptop, desktop, or VM)—WSL2 is not supported as an LXD host; please use native Ubuntu or an Ubuntu VM—internet access, a Google account, and a Telegram account.\r\n\r\nMore info: https:\u002F\u002Fevents.canonical.com\u002Fevent\u002F146\u002Fcontributions\u002F931\u002F",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FVTFUFD",{"id":6394,"room":6395,"start":3465,"end":4830,"language":96,"track":6396,"speakers":6398,"zh":6402,"en":6405,"tags":6406,"uri":6407},"VV9GT7",{"en":3731,"zh-hant":10},{"id":3734,"name":6397},{"en":3736,"zh-hant":3736},[6399],{"id":3915,"avatar":3916,"zh":6400,"en":6401},{"name":3918,"bio":3919},{"name":3918,"bio":3919},{"title":6403,"describe":6404,"type":30},"Building Resilient Hybrid Cloud-Native Architectures for Edge Inference","The intersection of cloud-native principles and edge computing necessitates robust, scalable architectures capable of handling intermittent connectivity and resource constraints. This talk explores the design patterns for building resilient hybrid infrastructure tailored for local inference tasks. We will discuss the challenges of maintaining state and orchestrating services across distributed environments using Kubernetes. The session will detail how to architect control planes that gracefully handle edge node disconnections and the strategies for deploying containerized AI workloads seamlessly across diverse hardware environments. By focusing on the foundational cloud-native technologies, attendees will learn how to construct reliable systems that bring computation directly to the user's environment.",{"title":6403,"describe":6404,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FVV9GT7",{"id":6409,"room":6410,"start":3505,"end":3709,"language":96,"track":6411,"speakers":6413,"zh":6417,"en":6420,"tags":6421,"uri":6422},"VZEBZ3",{"en":350,"zh-hant":10},{"id":3490,"name":6412},{"en":3492,"zh-hant":3493},[6414],{"id":6040,"avatar":6041,"zh":6415,"en":6416},{"name":6043,"bio":6044},{"name":6043,"bio":6044},{"title":6418,"describe":6419,"type":30},"我在農業現場: 2206永續專案窗口心得分享","農業數位與淨零轉型的現在進行式，由農業開放資料社群負責人阿柴現身說法，以上市櫃公司永續管理師的角度分享：企業如何規劃永續議題、跨域結合農業的機會、實務推動永續專案的心得。歡迎永續小白、尋求ＥＳＧ合作的農業人、以及單純想知道永續話題的聽眾入場交流。",{"title":6418,"describe":6419,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FVZEBZ3",{"id":6424,"room":6425,"start":3643,"end":3549,"language":328,"track":6426,"speakers":6428,"zh":6432,"en":6435,"tags":6436,"uri":6437},"W3WNAC",{"en":522,"zh-hant":10},{"id":3593,"name":6427},{"en":3595,"zh-hant":10},[6429],{"id":4018,"avatar":4019,"zh":6430,"en":6431},{"name":4021,"bio":4022},{"name":4021,"bio":4022},{"title":6433,"describe":6434,"type":30},"Revolution in Chip Design - Open Source RISC-V","Can a community design a robust open source RISC-V Core? Maybe even a whole MCU? Open Source, Verified so you can make a commercial Chip? That is what the OpenHW Foundation based in Europe is doing with it's members spanning the whole globe. Let us give you an update how our fully open source MCU got radiation hardened in a research project and who international companies use and contribute to our Cores like the famous CVA6 and what makes the Model work. Finally a quick overview about the different IP (including AI Accelerators) the OpenHW is hosting under an Apache 2.0 license.",{"title":6433,"describe":6434,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FW3WNAC",{"id":6439,"room":6440,"start":3668,"end":4013,"language":70,"track":6441,"speakers":6443,"zh":6447,"en":6450,"tags":6451,"uri":6452},"W9PG8F",{"en":326,"zh-hant":10},{"id":3268,"name":6442},{"en":3270,"zh-hant":3271},[6444],{"id":930,"avatar":931,"zh":6445,"en":6446},{"name":933,"bio":934},{"name":933,"bio":934},{"title":6448,"describe":6449,"type":30},"Building an MCP Server Using Postgres","Large language models are good at generating text, but can't easily report discrete, dynamic, and always-accurate results. They also can't take external actions. The model context protocol (MCP) has become the standard way for generative AI models to access external data and take external actions. This talk explains the purpose of MCPs, how they are used, and shows examples of database and non-database MCP servers.",{"title":6448,"describe":6449,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FW9PG8F",{"id":6454,"room":6455,"start":3549,"end":5287,"language":70,"track":6456,"speakers":6458,"zh":6466,"en":6469,"tags":6472,"uri":6473},"WAKGQH",{"en":326,"zh-hant":10},{"id":3268,"name":6457},{"en":3270,"zh-hant":3271},[6459],{"id":6460,"avatar":6461,"zh":6462,"en":6465},"YFWTUW","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FPXL_20210320_023758096.MP_ZOINVEy.jpg",{"name":6463,"bio":6464},"Amanda Lam","作為 Women Techmakers Hong Kong 大使及 ProductTank Hong Kong 活動召集人，過去 16 年 Amanda 每週也會共同主持其中一個最受歡迎的廣東話科技播客節目 - HKPUG Podcast《派樂派對》。日常方面，她以管理香港迪士尼樂園的數碼產品及平台來維持生計。 \r\n\r\nAs a Women Techmakers Hong Kong Ambassador and ProductTank Hong Kong Organizer, Amanda also co-hosts one of the most popular Cantonese tech podcasts, HKPUG Podcast for around 16 years. She manages digital products and platforms in Hong Kong Disneyland as her day job.",{"name":6463,"bio":6464},{"title":6467,"describe":6468,"type":30},"給 OpenClaw（龍蝦）控制你的 Home Assistant 智能家居，會發生甚麼事？","究竟讓 OpenClaw（龍蝦）利用 本地開放模型透過 Home Assistant 控制智能家居，會發生甚麼事？Amanda 會分享她的發現：究竟有甚麼更智能的自動化機會可以實踐？有甚麼方面不太可行？又發現了甚麼風險？\r\n\r\n這是 Amanda 在過往 Women Techmakers Hong Kong 國際婦女節活動、香港開源年會、OpenTechConf 及 COSCUP 開源人年會所做的 Home Assistant 系列分享的續篇。\r\n\r\n關鍵詞：#OpenClaw #LocalLLM #Gemma #Qwen #HomeAssistant #SmartHome #AgenticAI #AIagents #MCP",{"title":6470,"describe":6471,"type":30},"What happens when OpenClaw takes control to your Home Assistant smart home?","Amanda will discuss her discoveries to allow OpenClaw to control a Smart Home through Home Assistant via a local open LLM - what intelligent automation opportunities did it unlock, what didn't work well, and what risks were observed.\r\n\r\nThis is a sequel of the Home Assistant sharing series that Amanda previously shared at Women Techmakers Hong Kong IWD event, Hong Kong Open Source Conference, OpenTechConf and COSCUP Taipei.\r\n\r\nTags: #OpenClaw #LocalLLM #Gemma #Qwen #HomeAssistant #SmartHome #AgenticAI #AIagents #MCP",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FWAKGQH",{"id":6475,"room":6476,"start":4536,"end":4054,"language":96,"track":6477,"speakers":6479,"zh":6487,"en":6490,"tags":6492,"uri":6493},"WDF9UG",{"en":9,"zh-hant":10},{"id":3645,"name":6478},{"en":3647,"zh-hant":3648},[6480],{"id":6481,"avatar":6482,"zh":6483,"en":6486},"FSMLVN","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002F50255359392_e2932f8465_6k_L34ju8h_1_nufy1na.jpg",{"name":6484,"bio":6485},"Ko Ko","Ko Ko 連續當選多年微軟 AI 領域的最有價值專家（Microsoft AI MVP），曾任職於財星 500 強企業、美國矽谷科技新創、澳洲雪梨科技新創、美國上市公司等企業，擅長於 AI 技術與軟體開發。他也是大型技術年會的知名講者，包含了 ModernWeb、. NET Conference、PyCon APAC、PyCon HK、DevDays Asia 等等的國際大型年會，累計已超過萬名聽眾。他是技術社群 Chatbot Developers Taiwan 的核心成員，也經營粉專「大魔術熊貓工程師」。目前已經出版五本生成式 AI 應用的書籍。",{"name":6484,"bio":6485},{"title":6488,"describe":6489,"type":30},"Is Microsoft Agent Framework \"LangGraph for Enterprise\"? A Source Code Deep Dive","Microsoft Agent Framework 是 Microsoft 在 Semantic Kernel 與 AutoGen 之後推出的開源 AI Agent Framework，結合 Agent、Tool Calling、Handoff Orchestration 與 Graph-based Workflow Runtime 等設計。本次演講將以 Microsoft Agent Framework 為主角，深入剖析其原始碼，觀察 Agent 如何包裝 instructions、tools 與 model client，Handoff 如何轉移控制權，以及 WorkflowBuilder、Executor、Edge、Runner、Checkpointing 如何支撐多步驟與多 Agent 協作。OpenAI Agents SDK 與 LangGraph 將作為參照，用來比較 model-driven agent abstraction 與 graph workflow runtime 的設計差異，進一步討論 Microsoft Agent Framework 是否可被視為 \"LangGraph for Enterprise\"，以及它在開源 AI Agent Framework 生態中的設計定位。",{"title":6488,"describe":6491,"type":30},"Microsoft Agent Framework is Microsoft's open-source AI agent framework following Semantic Kernel and AutoGen, combining agents, tool calling, handoff orchestration, and graph-based workflow runtime design. This talk focuses on Microsoft Agent Framework as the main subject and examines its source code: how agents wrap instructions, tools, and model clients; how handoff transfers control; and how WorkflowBuilder, Executor, Edge, Runner, and checkpointing support multi-step and multi-agent collaboration. OpenAI Agents SDK and LangGraph will be used as reference points to compare model-driven agent abstractions with graph workflow runtimes, and to discuss whether Microsoft Agent Framework can be understood as \"LangGraph for Enterprise\" in the open-source AI agent ecosystem.",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FWDF9UG",{"id":6495,"room":6496,"start":3548,"end":3464,"language":70,"track":6497,"speakers":6499,"zh":6507,"en":6510,"tags":6511,"uri":6512},"WK9ZBQ",{"en":121,"zh-hant":10},{"id":4034,"name":6498},{"en":4036,"zh-hant":4036},[6500],{"id":6501,"avatar":6502,"zh":6503,"en":6506},"9XVENR","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002F9XVENR_vrrBwsc.webp",{"name":6504,"bio":6505},"Hong Minhee (洪 民憙)","I'm Hong Minhee (洪 民憙), a software engineer from Seoul. I write some free\u002Fopen source software, which are mostly written in Haskell, Python, and TypeScript. I'm an advocate of free\u002Fopen source software and [fediverse]. My professional interests are decentralized peer-to-peer networks and statically typed functional programming languages.\r\n\r\nI also have outside interests on etymology, East Asian languages (\u003Cabbr title=\"Chinese–Japanese–Korean\">CJK\u003C\u002Fabbr>), Chinese characters, and further, [Unihan].\r\n\r\nFor a more detailed introduction, please visit my personal website at \u003Chttps:\u002F\u002Fhongminhee.org\u002F>.\r\n\r\n[fediverse]: https:\u002F\u002Fwww.theverge.com\u002F24063290\u002Ffediverse-explained-activitypub-social-media-open-protocol\r\n[Unihan]: https:\u002F\u002Fwww.unicode.org\u002Freports\u002Ftr38\u002F",{"name":6504,"bio":6505},{"title":6508,"describe":6509,"type":30},"I just wanted ruby annotations: writing in dead scripts on the living fediverse","Koreans used to mix [hanja] (漢字; Chinese characters) into Korean prose, much as Japanese still mixes kanji and kana. The style is called [Korean mixed script] (國漢文混用體). Almost nobody writes this way anymore. I do.\r\n\r\nWhen I wanted to post this way on the fediverse, I ran into a small but stubborn problem: [`\u003Cruby>` annotations][ruby], the HTML feature that puts pronunciation guides above or beside characters, were stripped by the major servers I tried. I filed an issue against [Mastodon]. It sat there for a long time. At some point, “maybe I should run my own server” somehow became “maybe I should implement [ActivityPub] myself.”\r\n\r\nActivityPub is not simple. [JSON-LD] alone has several ways to say the same thing. Then come HTTP Signatures, WebFinger, NodeInfo, inbox forwarding, and the small incompatibilities that only become obvious when Mastodon and Misskey disagree. Before building the server I wanted, I built [Fedify]: a TypeScript framework that keeps most of that protocol plumbing out of application code.\r\n\r\n[Hollo] came next, because I still wanted the original thing: a single-user ActivityPub server where Markdown and `\u003Cruby>` annotations survived the trip. [Hackers' Pub] followed from the same framework, aimed at developers who want short posts and longer articles to federate.\r\n\r\nThis talk is about how a small typographic itch turned into upstream patches, a framework, and two fediverse servers. I still just wanted ruby annotations.\r\n\r\n[hanja]: https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FHanja\r\n[Korean mixed script]: https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FKorean_mixed_script\r\n[ruby]: https:\u002F\u002Fdeveloper.mozilla.org\u002Fen-US\u002Fdocs\u002FWeb\u002FHTML\u002FReference\u002FElements\u002Fruby\r\n[Mastodon]: https:\u002F\u002Fjoinmastodon.org\u002F\r\n[ActivityPub]: https:\u002F\u002Fwww.w3.org\u002FTR\u002Factivitypub\u002F\r\n[JSON-LD]: https:\u002F\u002Fjson-ld.org\u002F\r\n[Fedify]: https:\u002F\u002Ffedify.dev\u002F\r\n[Hollo]: https:\u002F\u002Fdocs.hollo.social\u002F\r\n[Hackers' Pub]: https:\u002F\u002Fhackers.pub\u002F",{"title":6508,"describe":6509,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FWK9ZBQ",{"id":6514,"room":6515,"start":3567,"end":4140,"language":96,"track":6516,"speakers":6518,"zh":6522,"en":6525,"tags":6526,"uri":6527},"WWMJMD",{"en":350,"zh-hant":10},{"id":3490,"name":6517},{"en":3492,"zh-hant":3493},[6519],{"id":6040,"avatar":6041,"zh":6520,"en":6521},{"name":6043,"bio":6044},{"name":6043,"bio":6044},{"title":6523,"describe":6524,"type":30},"我在農業現場-小微企業掏錢買SaaS產品的原因:以力MantaGO的台日拓展經驗","是什麼讓農業小老闆們願意花錢購買MantaGO的對話式商務平台?\r\n深耕小微企業七年多的女性創業者 Janice，如何看公司下一個七年的發展?\r\n\r\n以力科技成立於2018年，致力於為中小型企業提供簡單易上手的AI solution，透過技術解決資訊不對稱的問題。旗下品牌MantaGO對話式商務平台，以多渠道底層架構為基礎，整合5大社群平台如LINE、Messenger、Instagram、Google Business Chat及官網客服系統，提供市場完整的一站式商務服務。2025年以力科技在日本成立分公司，2026年更進一步推廣至泰國、越南等東南亞國家，協助更多海外品牌架構專屬AI聊天機器人，協助企業進行使用者管理優化，同時提升終端消費者體驗。\r\n\r\n對農業數位轉型有興趣的你，一定不能錯過他的分享！",{"title":6523,"describe":6524,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FWWMJMD",{"id":6529,"room":6530,"start":3910,"end":4700,"language":328,"track":6531,"speakers":6533,"zh":6541,"en":6544,"tags":6545,"uri":6546},"XA8PA8",{"en":899,"zh-hant":10},{"id":330,"name":6532},{"en":332,"zh-hant":332},[6534],{"id":6535,"avatar":6536,"zh":6537,"en":6540},"HPUVQL","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FHPUVQL_vIhqgmv.webp",{"name":6538,"bio":6539},"Aryan Kaushik","Aryan Kaushik is a GNOME Foundation and Ubuntu member, a member of the GNOME Internship Committee, and UbuCon Asia organising committee. He was a former GSoC’22 Intern at GNOME, GSoC'23 Intern under Chromium. Aryan was a speaker at GUADEC'22\u002F23\u002F25 and GNOME Asia'22-25 and was the lead local team organiser for UbuCon Asia 2024 in Jaipur, India and ex-lead of Ubuntu India.",{"name":6538,"bio":6539},{"title":6542,"describe":6543,"type":30},"From Community Need to Solution: The Story of Open Forms","Open Forms is a Linux application designed to provide a native, accessible, and offline-first form experience on Linux systems.\r\n\r\nAcross Ubuntu communities, LoCos, and open-source events, form-based workflows still rely heavily on proprietary web platforms. These solutions often fall short in areas critical to our ecosystem - offline usability, data ownership, accessibility in low-connectivity environments, and seamless integration with the Linux desktop.\r\n\r\nOpen Forms emerged from real-world needs within the Ubuntu community. It aims to provide a simple, reliable alternative that aligns with open-source values and Ubuntu’s vision of accessibility for everyone. And did I mention? It is available in a Snap!\r\n\r\nOpen Forms:\r\n\r\n - Enables offline-first data collection for booths, registrations, and surveys.\r\n- Integrates naturally with the Ubuntu desktop experience\r\n - Supports LoCos, community events, and grassroots initiatives\r\n- Runs efficiently on Linux mobile devices without requiring constant connectivity\r\n\r\nIn this talk, we will cover:\r\n\r\n - Why GTK4 + libadwaita were chosen for building a modern Ubuntu-native experience Key challenges and lessons learned while developing Open\r\n- Forms Real-world usage in Ubuntu booths and community events \r\n- How tools like Open Forms can strengthen Ubuntu communities across regions\r\n\r\nMore info: https:\u002F\u002Fevents.canonical.com\u002Fevent\u002F146\u002Fcontributions\u002F936\u002F",{"title":6542,"describe":6543,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FXA8PA8",{"id":6548,"room":6549,"start":6550,"end":6551,"language":328,"track":6552,"speakers":6554,"zh":6558,"en":6561,"tags":6562,"uri":6563},"XFYCTY",{"en":1253,"zh-hant":10},"2026-08-09T16:15:00+08:00","2026-08-09T17:00:00+08:00",{"id":1258,"name":6553},{"en":1260,"zh-hant":1261},[6555],{"id":1264,"avatar":1265,"zh":6556,"en":6557},{"name":1267,"bio":10},{"name":1267,"bio":10},{"title":6559,"describe":6560,"type":30},"Closing X ⚡Lightning talk \u002F 閃電秀 ⚡","閃電秀",{"title":6559,"describe":6560,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FXFYCTY",{"id":6565,"room":6566,"start":3667,"end":3668,"language":70,"track":6567,"speakers":6569,"zh":6577,"en":6580,"tags":6581,"uri":6582},"XLKBSU",{"en":1154,"zh-hant":10},{"id":286,"name":6568},{"en":288,"zh-hant":288},[6570],{"id":6571,"avatar":6572,"zh":6573,"en":6576},"XZ7VJE","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FXZ7VJE_gLrePtP.webp",{"name":6574,"bio":6575},"Julien Rouhaud","Julien is a PostgreSQL Major Contributor. He has been working with PostgreSQL since verson 8.3, as a software engineer and a DBA. He's also the author of other open source projects such as PoWA (real-time tuning) and HypoPG (hypothetical indexes).\r\n\r\nHe’s also a former -treasurer of the PostgreSQL Europe association.",{"name":6574,"bio":6575},{"title":6578,"describe":6579,"type":30},"Index best practice and hypothetical indexes in PostgreSQL","Covering indexes, partial indexes, access methods, operator classes, Index-only scans... Indexes in PostgreSQL are extensible and have a lot of features, so it's not always easy to know how to best use them or when not to use them.\r\n\r\nIn this talk we will cover multiple areas of indexing, giving example scenario example and best practice to highlight their use cases, and general improvements around indexes in the recent releases.\r\n\r\nWe will also cover some extensions around indexing, including hypopg that allows you to create hypothetical indexes to help you decide whether an index should be created or not, and other external tools to detect various problems around indexes.",{"title":6578,"describe":6579,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FXLKBSU",{"id":6584,"room":6585,"start":3968,"end":3969,"language":96,"track":6586,"speakers":6588,"zh":6596,"en":6599,"tags":6600,"uri":6601},"XLMBG9",{"en":39,"zh-hant":10},{"id":384,"name":6587},{"en":386,"zh-hant":10},[6589],{"id":6590,"avatar":6591,"zh":6592,"en":6595},"EUWGBF","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002F2023-04-17_17.00.47_8VhhHqr.jpg",{"name":6593,"bio":6594},"Ping Chen","軟體工程師，分享一些區塊鏈新科技之類的東東",{"name":6593,"bio":6594},{"title":6597,"describe":6598,"type":30},"Token 是硬幣，也是憑證：基於區塊鏈的聲譽系統的新趨勢","「Token」既可以翻譯成硬幣，也可以翻譯成憑證。過去幾年，區塊鏈社群反覆嘗試把人的聲譽、影響力與信任關係代幣化，但早期專案大多沒有成功：個人代幣常被炒作成網紅迷因幣；去中心化聲譽平臺則難以和既有社群網站競爭，因產品過重、缺乏 traction 而沒落。\r\n\r\n我們會從過去的失敗案例出發，討論近期幾個新的技術與產品方向，包括：零知識證明的身份，將不同來源的信任「加總」在一起使用、從應用中抽離出來的社交圖譜、讓聲譽能像貨幣一樣傳導的網路設計，以及 EIP-8004 所描繪的 Agent 之間鏈上支付並評價的情境。當未來的交易、評價與身份訊號逐漸可被鏈上觀察，我們也可能用類似反洗錢的 heuristic，辨識協同洗評價的帳號集群，在去中心化網路建立關於聲譽的新概念。",{"title":6597,"describe":6598,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FXLMBG9",{"id":6603,"room":6604,"start":3754,"end":3755,"language":96,"track":6605,"speakers":6607,"zh":6615,"en":6618,"tags":6621,"uri":6622},"YDJLTF",{"en":237,"zh-hant":10},{"id":241,"name":6606},{"en":243,"zh-hant":243},[6608],{"id":6609,"avatar":6610,"zh":6611,"en":6614},"GHWPCT","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FGHWPCT_timf9Ay.webp",{"name":6612,"bio":6613},"Meng-Ying Tsai (Fumi)","又名文月，簡稱 Fumi。現職為 PicCollage 的後端工程師，在寫一個叫 Ruby 的語言。\r\n喜歡各種亮亮的東西，前世可能是烏鴉。嗜好各種含有咖啡因的食物，像是茶、巧克力跟咖啡。不論是技術相關的東西，還是哪家咖啡廳很讚，都歡迎找我聊聊歐！",{"name":6612,"bio":6613},{"title":6616,"describe":6617,"type":30},"Locale 何處尋：從 Accept-Language 到 UGC","在這個議程中你將遇到...\r\n🎭 可愛又迷人的反派角色 Accept-Language\r\n✉️ 叫 mobile dev 傳 locale，卻收到謎之怪值...\r\n🧑 反派大魔王 UGC",{"title":6619,"describe":6620,"type":30},"How to handle locale: From Accept-Language to UGC","This talk explores:\r\n* What you should be aware of Accept-Language\r\n* Why you cannot always trust locale string from mobile app\r\n* How do people handle UGC in terms of localization",[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FYDJLTF",{"id":6624,"room":6625,"start":3504,"end":3505,"language":328,"track":6626,"speakers":6628,"zh":6632,"en":6635,"tags":6636,"uri":6637},"YTHB9R",{"en":522,"zh-hant":10},{"id":3593,"name":6627},{"en":3595,"zh-hant":10},[6629],{"id":175,"avatar":176,"zh":6630,"en":6631},{"name":178,"bio":179},{"name":178,"bio":179},{"title":6633,"describe":6634,"type":30},"Open Source accelerator chip and how RISC-V plays a role","Nekko open sourced the CORE-ET AI inference accelerator design based on Esperanto Technologies' ET-SOC-1 processor. This talk goes into how open sourcing RTL enables downstream  development, any what's now possible only on open source hardware and the RISC-V open ISA \r\n\r\nSource: https:\u002F\u002Fgithub.com\u002Fopenhwgroup\u002Fcore-et",{"title":6633,"describe":6634,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FYTHB9R",{"id":6639,"room":6640,"start":3732,"end":3991,"language":70,"track":6641,"speakers":6643,"zh":6651,"en":6654,"tags":6655,"uri":6656},"YWGJRR",{"en":93,"zh-hant":10},{"id":3467,"name":6642},{"en":3469,"zh-hant":3469},[6644],{"id":6645,"avatar":6646,"zh":6647,"en":6650},"YJGLDG","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FYJGLDG_mUfyZam.webp",{"name":6648,"bio":6649},"Hiroyuki Ishii","Hiroyuki Ishii is a Senior Architect at Panasonic Automotive Systems with over 10 years of experience in Linux-based automotive software platform. He began contributing to open source in 2021 when he joined the Automotive Grade Linux (AGL) project as a member of the Steering Committee and the System Architecture Team.\r\nHe is also active as a Linux Foundation Japan Evangelist, supporting open source collaboration and community growth across Japan, and helping bridge industry and OSS communities.",{"name":6648,"bio":6649},{"title":6652,"describe":6653,"type":30},"Automotive Grade Linux: An Open Source Platform Evolving Toward Software-Defined Systems","Automotive Grade Linux (AGL) is a collaborative open source project hosted by the Linux Foundation, building a common Linux-based software platform for automotive systems. This session introduces AGL through its recent evolution. After a brief overview of the project and its community, the talk highlights key updates from the past year to show how AGL is expanding beyond its original scope.\r\n\r\nTwo initiatives illustrate this direction. SoDeV (Software Defined Vehicle reference platform) explores software-first development through virtualization and hardware abstraction, improving portability and enabling experimentation across environments by leveraging promising open source projects such as Xen, Zephyr, and VirtIO. AGL Assessment Automation (AAA) applies modern open source security and supply-chain practices—originating in the cloud-native ecosystem—to large and complex automotive supply chains, addressing challenges common across embedded systems industries worldwide.\r\n\r\nRather than focusing on industry-specific use cases, the session emphasizes the technical ideas and open source practices behind these efforts, and concludes with a look at where AGL is heading and how developers can get involved.",{"title":6652,"describe":6653,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FYWGJRR",{"id":6658,"room":6659,"start":5465,"end":6660,"language":70,"track":6661,"speakers":6663,"zh":6671,"en":6674,"tags":6675,"uri":6676},"ZCCDSX",{"en":1154,"zh-hant":10},"2026-08-09T12:45:00+08:00",{"id":286,"name":6662},{"en":288,"zh-hant":288},[6664],{"id":6665,"avatar":6666,"zh":6667,"en":6670},"GZ9SFT","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FGZ9SFT_SaJp5k0.webp",{"name":6668,"bio":6669},"Robert Treat","Working on database-backed, internet-based systems for over a decade, Robert is a published author and long-time open source contributor, having been recognized as a Major Contributor to the PostgreSQL project for his work over the years. An international speaker on databases, open source, and managing web operations at scale, you can find him online at https:\u002F\u002Fpgtreats.info.",{"name":6668,"bio":6669},{"title":6672,"describe":6673,"type":30},"I Didn't Know Postgres Could Do That!","PostgreSQL is often thought of as \"The Database for DBAs\", but what more and more developers are finding out is that what makes Postgres \"The Worlds Most Advanced Open Source Database\" is its wide array of features, many of which are really geared towards users who want to build applications on top of Postgres.\r\n\r\nIn this talk, we'll look at some of the features in Postgres you may not have seen in other database systems, and talk about how you can make use of those features in your applications. Queries, indexing, data types, and more will all be up for discussion as we show you examples of just what Postgres can do.",{"title":6672,"describe":6673,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FZCCDSX",{"id":6678,"room":6679,"start":3567,"end":3568,"language":96,"track":6680,"speakers":6682,"zh":6686,"en":6689,"tags":6690,"uri":6691},"ZESBXN",{"en":326,"zh-hant":10},{"id":3268,"name":6681},{"en":3270,"zh-hant":3271},[6683],{"id":5197,"avatar":5198,"zh":6684,"en":6685},{"name":5200,"bio":5201},{"name":5200,"bio":5201},{"title":6687,"describe":6688,"type":30},"COCA MCP: Learning English While Chatting with an LLM","People who try to learn English as a second language usually quit their flashcard app within two weeks. The hard part isn't memorising definitions. It's that a new word only sticks after a dozen real-world encounters, and a flashcard never gives you that. We introduce the concept \"Ambient Learning\" — learning from natural conversation.\r\n\r\nCOCA MCP is an useful tool to suggest the vocabularies chat to an LLM. The server sits behind the Corpus of Contemporary American English — 60,000 words ranked by frequency — and exposes a small set of MCP tools. When a client like Claude is talking to a learner, it asks the server which words are due, and gets back a short list scheduled by FSRS-4.5. The model uses those words in its reply. It then tells the server which ones it actually used; the server records the exposure and reschedules.\r\n\r\nThe session walks through a modern English learning methodology on the running site at coca.gavinguo.cc — how to use MCP technology to improve the English ability.",{"title":6687,"describe":6688,"type":30},[62],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FZESBXN",{"id":6693,"room":6694,"start":3643,"end":3549,"language":70,"track":6695,"speakers":6697,"zh":6705,"en":6708,"tags":6709,"uri":6710},"ZQRGXM",{"en":192,"zh-hant":10},{"id":196,"name":6696},{"en":198,"zh-hant":198},[6698],{"id":6699,"avatar":6700,"zh":6701,"en":6704},"HXCYJC","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FHXCYJC_4VuehEL.webp",{"name":6702,"bio":6703},"Leesoo Ahn","Senior Software Engineer",{"name":6702,"bio":6703},{"title":6706,"describe":6707,"type":30},"Linux as a Guest OS: Apple Silicon's Virtualization Infrastructure","This session provides an {overview, in-depth} architectural breakdown of running Linux Kernel within Apple Silicon’s Virtualization Infrastructure. We will explore it from the ground up, examining what the _Hypervisor.framework_ and _Virtualization.framework_ are.\r\n\r\nThe talk will specifically cover:\r\n- An Introduction to Apple Silicon's Virtualization Stacks\r\n- Hardware-Assisted CPU Emulation\r\n- VirtIO: Peripheral Interfaces for VMs\r\n- User-Space Utilities for Virtualization",{"title":6706,"describe":6707,"type":30},[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FZQRGXM",{"id":6712,"room":6713,"start":3754,"end":3755,"language":96,"track":6714,"speakers":6716,"zh":6724,"en":6727,"tags":6729,"uri":6730},"ZV88PP",{"en":263,"zh-hant":10},{"id":44,"name":6715},{"en":46,"zh-hant":47},[6717],{"id":6718,"avatar":6719,"zh":6720,"en":6723},"FAGUY7","https:\u002F\u002Fpretalx.coscup.org\u002Fmedia\u002Favatars\u002FFAGUY7_sDACwwd.webp",{"name":6721,"bio":6722},"Yusef Schultz","I'm Yusef",{"name":6721,"bio":6722},{"title":6725,"describe":6726,"type":30},"區塊鏈網路上基於開放標準的實體身份識別方法","世界各地因為區塊鏈洗錢的問題頻繁導致各國政府開始參考並引入區塊鏈防洗錢法案避免國際犯罪，他們的共同特色就是定位使用者本身。 而台灣稱為虛擬資產服務法已在 2026 年 4 月 2 日通過，台灣人的身份被如何定位也將逐漸會成為台灣虛擬資產持有者關注的議題。\r\n\r\n本次議題將會分享現代大家都用了哪些開放的標準去實現身份定位，以及討論密碼學術界對區塊鏈產業帶來的正面技術影響。",{"title":6728,"describe":6726,"type":30},"Open Standard and Identification Methodology in the Blockchain",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FZV88PP",{"id":6732,"room":6733,"start":3505,"end":3732,"language":96,"track":6734,"speakers":6736,"zh":6740,"en":6743,"tags":6746,"uri":6747},"ZZNCJ9",{"en":282,"zh-hant":10},{"id":3670,"name":6735},{"en":3672,"zh-hant":3673},[6737],{"id":5485,"avatar":5486,"zh":6738,"en":6739},{"name":5488,"bio":5489},{"name":5488,"bio":5489},{"title":6741,"describe":6742,"type":30},"我用 OWASP Top 10 掃了 Gemma 4，這是成績單","Google 推出 Gemma 4 的時候附了 model card 和 safety benchmark，但那些數字是 Google 自己測的。如果你用 OWASP LLM Top 10 的標準，從攻擊者視角重新掃一遍，結果會一樣嗎？\r\n這場議程分享我們在 NVIDIA DGX Spark GB10 和 Mac mini M4（MLX）上對 Gemma 4 26B 進行紅隊掃描的完整過程與結果。我們用 OpenClaw 攻擊鏈逐項測試 OWASP LLM Top 10 的風險項目，包括 Prompt Injection（LLM01：7 項測試通過 5 項）、Visual Prompt Injection（3 項測試確認 2 項弱點），以及其他類別的實測數據。\r\n你會看到的不是理論分析，而是一份有具體 pass\u002Ffail 的成績單。我們會拆解每一項測試的攻擊手法、Gemma 4 的實際反應、以及哪些弱點可以用 NemoClaw Guardrail 補強。\r\n這場的目標不是說 Gemma 4 不安全，而是讓每一個用開源模型的開發者知道：model card 上沒寫的風險，你要自己測。我們會開源測試腳本，讓你對任何開源 LLM 跑同樣的弱掃流程。",{"title":6744,"describe":6745,"type":30},"I Scanned Gemma 4 Against OWASP Top 10 — Here's the Report Card","When Google released Gemma 4, it came with a model card and safety benchmarks — tested by Google themselves. But what happens when you rescan it from an attacker's perspective using OWASP LLM Top 10 criteria?\r\nThis talk shares the full process and results of our red team scan on Gemma 4 26B, run on an NVIDIA DGX Spark GB10 and Mac mini M4 (MLX). We used the OpenClaw attack chain to test each OWASP LLM Top 10 risk category, including Prompt Injection (LLM01: 5 out of 7 tests passed), Visual Prompt Injection (2 out of 3 confirmed vulnerabilities), and real test data across other categories.\r\nThis is not a theoretical analysis — it's a report card with concrete pass\u002Ffail results. We'll walk through the attack techniques used in each test, how Gemma 4 actually responded, and which weaknesses can be mitigated with NemoClaw Guardrail.\r\nThe goal isn't to say Gemma 4 is insecure. It's to show every developer using open-source models that the risks not listed on the model card are yours to test. We'll open-source our testing scripts so you can run the same vulnerability scan on any open-source LLM.",[88],"https:\u002F\u002Fcoscup.org\u002F2026\u002Fsession\u002FZZNCJ9",1783820807114]