綜合議程 - 各種開源議題

WrenAI 介紹

WrenAI 是你的 AI 數據助手，幫助你無需撰寫 SQL 就能更快地獲取結果和見解，而且它是開源的！

為什麼我們開發了 WrenAI？

我們在為客戶提供服務時發現，數據團隊不斷收到臨時數據需求會導致一系列問題，比如專注力和生產力下降、壓力和倦怠感增加，以及資源利用效率低下。

為什麼 WrenAI 特別？

WrenAI 想要落實 RAG Text-to-SQL 的架構在開源社群上，讓這項技術可以普及到各個國家各行各業的使用者，幫助加速數據轉型與提升企業生產力，以下為我們的三大獨特設計：

• Semantics modeling & Indexing (索引): WrenAI 獨特設計了 "Modeling Definition Language"(MDL) 能夠讓 LLM 去理解資料結構以及企業內部的語意，讓 LLM 更貼近企業的營運情境，讓使用者問與答更精準。
• Augmentation (增強): 透過 WrenAI 的 MDL 設計，去強化LLM的 Text-to-SQL 能力讓 LLM 能夠增強上下文的理解能力，能夠有效的降低幻覺出現。
• Generation (生成): WrenAI SQL 生成的過程中提供使用者來回確認的流程設計，以及在 SQL 產生後進行 SQL 驗證與校正，讓 SQL 生成更精準。

未來功能規劃（僅列舉部分）

• LLM系統釐清使用者詢問
• 使用者反饋
• LLM系統表現評估

講者

Jimmy Yeh

I am working on a problem that humans have been attempting to solve for 50 years.

Software Product Engineer in Canner

PMPC(Public Money, Public Code) is a concept promoted by EU since 2017. Which is a good for promoting the free/open-source culture, yet still leave some pitfalls which might be used to underturn the original ideal if not executed carefully. This talk gives a brief introduction of PMPC, followed by reviewing certain failed cases in Taiwan in the past 20 years and what still going wrong right now. Then I will propose a principle of PMPTDS, a more complete strategic model to address the previous issues for promoting FOSS or open data when involving public resources, and how it can actually mitigate the previous issues.

講者

Kuo-Chieh Ting

Kuo-Chieh got dual graduate degrees in Computer Science, and Art and Technology. He is a veteran of FOSS and Open Data movement. He is among the first generation Chinese users of Libreoffice and Mageia Linux（named Star Office/Open Office and Mandriva back then, respectively) He has been actively involved in Chinese language l10n community events. He has initiailed the Chinese Electronic Literature Library, which is a missing piece of the modern Chinese NLP which Taiwan needs to compete with other rivals in AI, and came up with the ideas of two big FOSS projects funded by the Taiwan government. Recently he shifts more efforts on focusing using FOSS tools and open data to create arts and community sharing ecology, such as leading high school students to create and share the 3D model of a historical architecture from scratch, during learning the FOSS 2D/3D tools and how engineers cooperate with each other in the FOSS community.

You picked a license for your project, but what now? What is best practice to add licensing information? How do you verify that you got it right? And what about tools that work with licensing information in an automated way?

These questions can be difficult, especially when reusing software from different projects that are released under various different licenses. But using code from multiple sources is what Software Freedom is about. This is where REUSE comes in: The Free Software Foundation Europe (FSFE) started REUSE to provide a set of recommendations to make licensing your Free Software projects easier. Not only do these recommendations make it easier for you to declare the licenses under which your works are released, they also make it easier for a computer to understand how your project is licensed. With three simple steps, you can follow best practice and you can verify that with the REUSE tool.

講者

Florian Snow

Florian Snow is a software developer from Nuremberg, Germany with a background in a variety of different fields such as teaching, programming, linguistics, and design. He became interested in Free Software around 1999 and has been active in the community since around 2006. Apart from his job as a software developer, he works and volunteers for the Free Software Foundation Europe where he has been active in many different areas over the years.

數位發展部數位產業署正全力推動台灣的AI轉型，推出數位公共建設計畫-軟體基盤，以利減少重複開發投資，並提升軟體數位服務品質和效率。首先我們提供免費AI算力資源，這些資源包括Nvidia H100和AMD MI300X虛擬化GPU，幫助資服產業充分利用資源進行創新！

開源的力量能夠加速領域應用開發，讓更多人參與到創新中來，為了促進開源社群的合作，我們將推出「程式碼信任共享平台」，這個平台促進程式碼和AI模型的資源共享，確保數位建設的安全性和合規性（Openchain）。

此外，為鼓勵企業進行研發創新，提供稅務抵減優惠來降低研發成本，激勵企業加大對數位科技和創新應用的投資，旨在支持企業進行研發創新，推動整個軟體生態系的發展。

數位產業署致力於建構開放、創新和安全的數位生態系，促進資訊產業的使用和發展。我們將與大家分享目前在軟體基盤正推動的相關業務，並期待和開源夥伴們共同推動開源技術在台灣的普及與應用!

講者

巫建恒

現職： 數位發展部數位產業署科長

經歷： 台北市政府衛生局資訊室主任、 主計總處主計資訊處分析師

學歷： 台灣科技大學資訊管理系博士、 台灣科技大學機械工程系碩士

Zenoh 是 Eclipse Foundation 下的開源軟體，有極佳通訊的效能，適合用在機器人和自駕車領域中各個節點之間的通訊。 目前此技術已經被 ROS 2、Autoware、Eclipse uProtocol 等專案採納。 本次分享會介紹 Zenoh 以及我們如何將其與其他開源社群合作並整合的經驗和故事。

講者

郭鎭頴

對開源軟體/社群極感興趣的工程師，對機器人、自駕車、通訊技術有些研究。目前主要是負責將 Zenoh 與其他開源軟體進行整合，非常熱愛工作與開源的結合。

As the wave of AI surges across the world, everyone is impressed by its performance and excited about leveraging its power to streamline daily tasks. At Canonical Ubuntu, we're actively developing a new AI-powered assistant that analyzes incoming cases, identifies potential causes, and suggests solutions based on existing knowledge sources. In this talk, Gavin will provide a comprehensive overview and discuss implementation ideas for the framework.

世界各地正迎來人工智能熱潮，人們對其表現讚嘆不已，並熱切期待利用其强大功能簡化日常工作。Canonical Ubuntu公司正積極開發一種新型的人工智能助理，該助理可以分析新收到的客戶問題，識別根本的問題所在，並且根據現有的知識庫提出解决方案。在本演講中，Gavin將介紹整個設計框架的概況並討論靈感的來源。

講者

Gavin Guo

Gavin Guo was a Linux kernel developer in the Ubuntu community. He was the speaker of Spectre v2 Internal at the 2018 China L3C and KASan debugging at the 2016 China Linux Kernel Conference. He worked for Canonical in the Customer Success division. He was responsible for the kernel stability and performance tuning of the OpenStack platform, especially in the areas of NUMA (Non­uniform Memory Access), Page Reclaim, and the SLUB allocator. He was the one who introduced KASAN into the team to investigate kernel issues on the OpenStack platform, which put an end to many nightmares.

Currently, Gavin is the engineering manager of the APAC sustaining engineering team at Canonical. He leads a distributed team in the Asia Pacific region, with engineers from diverse backgrounds, from user to kernel space. They work together to identify and solve complicated issues in large-scale environments, especially in performance evaluation, analysis, and tuning for the Ubuntu cloud.

This proposal presents a pioneering initiative to integrate OpenStack, an open-source cloud computing platform, with FreeBSD, a robust Unix-like operating system. Traditionally, OpenStack has been closely associated with Linux-based environments, leveraging specific Linux features and technologies. This integration aims to expand OpenStack’s applicability by harnessing FreeBSD’s advanced networking, security, and efficient resource management capabilities.

The project scope involves adapting OpenStack’s key components to function seamlessly within FreeBSD’s system architecture, focusing on virtualization with bhyve and FreeBSD’s unique networking stack. A Proof of Concept (PoC) has been successfully developed, demonstrating the viability of this integration and laying a foundation for further development.

We address several critical challenges in this integration process, including adapting libvirt for bhyve, managing VLANs, modifying Open vSwitch for FreeBSD, ensuring efficient DHCP services, and aligning FreeBSD’s network namespace and firewall functionalities with OpenStack’s requirements. The project also tackles the adaptation of OpenStack’s oslo.privsep library to FreeBSD’s privilege model and addresses the complexities of nested virtualization and VM console access within FreeBSD.

Future work involves expanding the integration to more OpenStack components, enhancing system performance, and fostering collaboration within the FreeBSD and OpenStack communities. This integration represents a significant advancement in cloud computing, offering a versatile platform that combines the strengths of both OpenStack and FreeBSD. The project invites collaboration and contribution from the community to overcome the challenges and fully realize the potential of this innovative integration.

講者

Zespre Chang

Hi, I'm Chih-Hsin Chang. Most of the time, people call me Zespre. I'm working at SUSE and developing an open-sourced HCI project called Harvester. Due to my previous experience in OpenStack and FreeBSD, I also worked on a side project that focused on porting essential OpenStack components from the Linux world to FreeBSD.

對我來說，Open Source 的意義在於賦能：任何人都可以拿他建立更好的東西。然而，傳統上這種能力主要限於會寫程式的工程師。隨著 AI 技術的發展，尤其是 ChatGPT 的出現，這種情況正在改變。現在，即使是沒有工程師背景的一般老百姓，也能通過 AI 生成程式碼，進而參與開源軟體專案。

演講目標 - 降低門檻：介紹如何利用 ChatGPT 等 AI 工具生成程式碼，降低非工程師參與開源專案的門檻。 - 實際應用：教會會眾一些實際的自動化解決方案，讓他們能夠解決自己的問題。 - 開源貢獻：引導會眾將他們編寫的自動化軟體以開源方式釋出，進一步推動開源社群的發展。

講者

李慕約

李慕約是 Generative AI 社群 ( gaiconf.com ) 的主辦人，這是一個致力於推廣生成式 AI 的社群。

李慕約同時也是李慕約公司（muyueh.com）的負責人，這家公司提供資料視覺化、大數據分析等服務，客戶包括中華電信、遠傳電信、玉山銀行等企業，

李慕約過去曾任 COSUCP 2014、2015、2016 的講者，講述開放資料以及資料視覺化的主題。

啟發自 Z80-MBC2 的微電腦製作，我將 Z80 與 SRAM 首先連接到麵包板上進行實驗。隨後，我深入研究了 Z80-MBC2 的原始碼，以理解其上傳原理。為了執行這一計劃，我選擇使用 Arduino Uno 作為載入器核心。

講者

鄭鴻旗

鄭鴻旗為 Openlab.Taipei 社群發起人者之一，以藝術學習為背景，也關注著科技、設計與藝術的結合與發展，在這研究的過程中遇見了自由和開放原始碼軟、硬體等相關科技與社群後，對於背後的精神、文化與哲學著迷，喜歡 Diy、Hack、Make、程式與數位製造相關事物，並且將推廣開源技術與藝術交流的社群視為己志。

講者

pondahai

我是彭大海

ActivityPub 是一個用來連結不同社群網站的協定，透過 ActivityPub，我們可以在不同的社群網站之間互相關注、互相轉發、互相回覆。這個協定的目標是讓使用者可以自由選擇社群網站，而不用擔心因為選擇了不同的社群網站而無法互相溝通。在這個演講中，我們將會介紹 ActivityPub 的基本概念，並且實際演示如何使用 ActivityPub 來建立一個簡單的社群網站。

講者

pichu chen

先前曾經參與過 PTT APP 後端開發的皮丘。

周末晚上會定期出現在 VRChat

Audio and video capabilities have become widely integrated into various industries, including e-commerce live streaming, educational live streaming, online meetings, and interactive entertainment, constantly changing the existing industry landscape.

However, building an audio and video platform for any scenario requires dealing with client-side to streaming media server, as well as various complex functions such as authentication, recording, forwarding, transcoding, virtual live streaming, protocol conversion, and low latency. This requires a comprehensive and easy-to-use out-of-the-box solution.

SRS Stack is a ready-to-use audio and video solution with an All-in-One design, allowing developers familiar with the business to quickly add audio and video capabilities to their existing services.

講者

Winlin Yang

Winlin is an accomplished software engineer and a distinguished leader in open-source projects, with an impressive track record of over 10+ years in video streaming and cloud industries. His expertise includes cloud service of RTC, new use scenarios of rich media, and the development of the widely used open source project, SRS (Simple Realtime Server).

Winlin works in Tencent Cloud, delivering innovative, low-latency interactive live streaming solutions for group audio/video calls. Before his current role, he worked at ChinaCache, successfully developing the live-streaming CDN server. He also built the WebRTC cloud service from scratch in Alibaba Cloud.

People widely recognize Winlin as the founder and driving force behind the success of the SRS (Simple Realtime Server) project. This powerful real-time video gateway server has been valued and widely used globally. With his expertise and leadership, he is truly impacting the world of video streaming and cloud technologies.

開放文化與資安管理是衝突的嗎？本議程將從工程師的角度出發，探索「開源／CC授權 vs 資安管理」的邊界。

講者

Denken Chen

軟體開發與寫作經驗超過十年。作品可見個人網站：https://denkeni.org

Diátaxis 是一種關於文件的思考和實施方式。

它根據系統地理解文件用戶的需求，規定了內容、架構和形式的處理方法。 Diátaxis 確定了四種不同的需求，以及四種相應的文件形式——教程、操作指南、技術參考和解釋。它將它們放在一個系統的關係中，並提出文件本身應圍繞這些需求的結構進行組織。

Diataxis 所服務的使用者是某個技能領域的從業人員. 技能領域是由工藝定義的 - 工具或產品的使用就是工藝 使用程式語言是一門手藝, 就像駕駛一輛特定的飛機, 甚至成為飛行員一樣.

Diataxis 解決了文件內容(寫什麼), 風格(如何寫), 和架構(如何組織)相關的問題

講者

jneo8

jneo8 is a engineer focus on cloud technology and love open source world.

長庚大學基於教學上的需要，使用 kubernetes+kubeflow 建立了一套雲端的 AI 教學系統。根據教學管理上的需要，客制化 kubeflow 並加入了下列的新模組。這個系統除了給一般的 AI 程式教學外，也很適合給一般分享 GPU 硬體給多人使用。以提高資源的移用率。

我們將介紹 kubeflow 的架構。並說明我們對 kubeflow 的修改，並討論將來一至二年的計畫。希望能邀請更多人加入共同開發出一套功能更完善的系統。

講者

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。