[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"similar-chiphuyen--aie-book":3,"tool-chiphuyen--aie-book":62},[4,18,28,37,45,53],{"id":5,"name":6,"github_repo":7,"description_zh":8,"stars":9,"difficulty_score":10,"last_commit_at":11,"category_tags":12,"status":17},4358,"openclaw","openclaw\u002Fopenclaw","OpenClaw 是一款专为个人打造的本地化 AI 助手，旨在让你在自己的设备上拥有完全可控的智能伙伴。它打破了传统 AI 助手局限于特定网页或应用的束缚，能够直接接入你日常使用的各类通讯渠道，包括微信、WhatsApp、Telegram、Discord、iMessage 等数十种平台。无论你在哪个聊天软件中发送消息，OpenClaw 都能即时响应，甚至支持在 macOS、iOS 和 Android 设备上进行语音交互，并提供实时的画布渲染功能供你操控。\n\n这款工具主要解决了用户对数据隐私、响应速度以及“始终在线”体验的需求。通过将 AI 部署在本地，用户无需依赖云端服务即可享受快速、私密的智能辅助，真正实现了“你的数据，你做主”。其独特的技术亮点在于强大的网关架构，将控制平面与核心助手分离，确保跨平台通信的流畅性与扩展性。\n\nOpenClaw 非常适合希望构建个性化工作流的技术爱好者、开发者，以及注重隐私保护且不愿被单一生态绑定的普通用户。只要具备基础的终端操作能力（支持 macOS、Linux 及 Windows WSL2），即可通过简单的命令行引导完成部署。如果你渴望拥有一个懂你",349277,3,"2026-04-06T06:32:30",[13,14,15,16],"Agent","开发框架","图像","数据工具","ready",{"id":19,"name":20,"github_repo":21,"description_zh":22,"stars":23,"difficulty_score":24,"last_commit_at":25,"category_tags":26,"status":17},9989,"n8n","n8n-io\u002Fn8n","n8n 是一款面向技术团队的公平代码（fair-code）工作流自动化平台，旨在让用户在享受低代码快速构建便利的同时，保留编写自定义代码的灵活性。它主要解决了传统自动化工具要么过于封闭难以扩展、要么完全依赖手写代码效率低下的痛点，帮助用户轻松连接 400 多种应用与服务，实现复杂业务流程的自动化。\n\nn8n 特别适合开发者、工程师以及具备一定技术背景的业务人员使用。其核心亮点在于“按需编码”：既可以通过直观的可视化界面拖拽节点搭建流程，也能随时插入 JavaScript 或 Python 代码、调用 npm 包来处理复杂逻辑。此外，n8n 原生集成了基于 LangChain 的 AI 能力，支持用户利用自有数据和模型构建智能体工作流。在部署方面，n8n 提供极高的自由度，支持完全自托管以保障数据隐私和控制权，也提供云端服务选项。凭借活跃的社区生态和数百个现成模板，n8n 让构建强大且可控的自动化系统变得简单高效。",184740,2,"2026-04-19T23:22:26",[16,14,13,15,27],"插件",{"id":29,"name":30,"github_repo":31,"description_zh":32,"stars":33,"difficulty_score":10,"last_commit_at":34,"category_tags":35,"status":17},10095,"AutoGPT","Significant-Gravitas\u002FAutoGPT","AutoGPT 是一个旨在让每个人都能轻松使用和构建 AI 的强大平台，核心功能是帮助用户创建、部署和管理能够自动执行复杂任务的连续型 AI 智能体。它解决了传统 AI 应用中需要频繁人工干预、难以自动化长流程工作的痛点，让用户只需设定目标，AI 即可自主规划步骤、调用工具并持续运行直至完成任务。\n\n无论是开发者、研究人员，还是希望提升工作效率的普通用户，都能从 AutoGPT 中受益。开发者可利用其低代码界面快速定制专属智能体；研究人员能基于开源架构探索多智能体协作机制；而非技术背景用户也可直接选用预置的智能体模板，立即投入实际工作场景。\n\nAutoGPT 的技术亮点在于其模块化“积木式”工作流设计——用户通过连接功能块即可构建复杂逻辑，每个块负责单一动作，灵活且易于调试。同时，平台支持本地自托管与云端部署两种模式，兼顾数据隐私与使用便捷性。配合完善的文档和一键安装脚本，即使是初次接触的用户也能在几分钟内启动自己的第一个 AI 智能体。AutoGPT 正致力于降低 AI 应用门槛，让人人都能成为 AI 的创造者与受益者。",183572,"2026-04-20T04:47:55",[13,36,27,14,15],"语言模型",{"id":38,"name":39,"github_repo":40,"description_zh":41,"stars":42,"difficulty_score":10,"last_commit_at":43,"category_tags":44,"status":17},3808,"stable-diffusion-webui","AUTOMATIC1111\u002Fstable-diffusion-webui","stable-diffusion-webui 是一个基于 Gradio 构建的网页版操作界面，旨在让用户能够轻松地在本地运行和使用强大的 Stable Diffusion 图像生成模型。它解决了原始模型依赖命令行、操作门槛高且功能分散的痛点，将复杂的 AI 绘图流程整合进一个直观易用的图形化平台。\n\n无论是希望快速上手的普通创作者、需要精细控制画面细节的设计师，还是想要深入探索模型潜力的开发者与研究人员，都能从中获益。其核心亮点在于极高的功能丰富度：不仅支持文生图、图生图、局部重绘（Inpainting）和外绘（Outpainting）等基础模式，还独创了注意力机制调整、提示词矩阵、负向提示词以及“高清修复”等高级功能。此外，它内置了 GFPGAN 和 CodeFormer 等人脸修复工具，支持多种神经网络放大算法，并允许用户通过插件系统无限扩展能力。即使是显存有限的设备，stable-diffusion-webui 也提供了相应的优化选项，让高质量的 AI 艺术创作变得触手可及。",162132,"2026-04-05T11:01:52",[14,15,13],{"id":46,"name":47,"github_repo":48,"description_zh":49,"stars":50,"difficulty_score":24,"last_commit_at":51,"category_tags":52,"status":17},1381,"everything-claude-code","affaan-m\u002Feverything-claude-code","everything-claude-code 是一套专为 AI 编程助手（如 Claude Code、Codex、Cursor 等）打造的高性能优化系统。它不仅仅是一组配置文件，而是一个经过长期实战打磨的完整框架，旨在解决 AI 代理在实际开发中面临的效率低下、记忆丢失、安全隐患及缺乏持续学习能力等核心痛点。\n\n通过引入技能模块化、直觉增强、记忆持久化机制以及内置的安全扫描功能，everything-claude-code 能显著提升 AI 在复杂任务中的表现，帮助开发者构建更稳定、更智能的生产级 AI 代理。其独特的“研究优先”开发理念和针对 Token 消耗的优化策略，使得模型响应更快、成本更低，同时有效防御潜在的攻击向量。\n\n这套工具特别适合软件开发者、AI 研究人员以及希望深度定制 AI 工作流的技术团队使用。无论您是在构建大型代码库，还是需要 AI 协助进行安全审计与自动化测试，everything-claude-code 都能提供强大的底层支持。作为一个曾荣获 Anthropic 黑客大奖的开源项目，它融合了多语言支持与丰富的实战钩子（hooks），让 AI 真正成长为懂上",161692,"2026-04-20T11:33:57",[14,13,36],{"id":54,"name":55,"github_repo":56,"description_zh":57,"stars":58,"difficulty_score":59,"last_commit_at":60,"category_tags":61,"status":17},8272,"opencode","anomalyco\u002Fopencode","OpenCode 是一款开源的 AI 编程助手（Coding Agent），旨在像一位智能搭档一样融入您的开发流程。它不仅仅是一个代码补全插件，而是一个能够理解项目上下文、自主规划任务并执行复杂编码操作的智能体。无论是生成全新功能、重构现有代码，还是排查难以定位的 Bug，OpenCode 都能通过自然语言交互高效完成，显著减少开发者在重复性劳动和上下文切换上的时间消耗。\n\n这款工具专为软件开发者、工程师及技术研究人员设计，特别适合希望利用大模型能力来提升编码效率、加速原型开发或处理遗留代码维护的专业人群。其核心亮点在于完全开源的架构，这意味着用户可以审查代码逻辑、自定义行为策略，甚至私有化部署以保障数据安全，彻底打破了传统闭源 AI 助手的“黑盒”限制。\n\n在技术体验上，OpenCode 提供了灵活的终端界面（Terminal UI）和正在测试中的桌面应用程序，支持 macOS、Windows 及 Linux 全平台。它兼容多种包管理工具，安装便捷，并能无缝集成到现有的开发环境中。无论您是追求极致控制权的资深极客，还是渴望提升产出的独立开发者，OpenCode 都提供了一个透明、可信",144296,1,"2026-04-16T14:50:03",[13,27],{"id":63,"github_repo":64,"name":65,"description_en":66,"description_zh":67,"ai_summary_zh":67,"readme_en":68,"readme_zh":69,"quickstart_zh":70,"use_case_zh":71,"hero_image_url":72,"owner_login":73,"owner_name":74,"owner_avatar_url":75,"owner_bio":76,"owner_company":77,"owner_location":78,"owner_email":77,"owner_twitter":79,"owner_website":80,"owner_url":81,"languages":82,"stars":87,"forks":88,"last_commit_at":89,"license":77,"difficulty_score":59,"env_os":90,"env_gpu":91,"env_ram":91,"env_deps":92,"category_tags":95,"github_topics":77,"view_count":24,"oss_zip_url":77,"oss_zip_packed_at":77,"status":17,"created_at":96,"updated_at":97,"faqs":98,"releases":99},10079,"chiphuyen\u002Faie-book","aie-book","[WIP] Resources for AI engineers. Also contains supporting materials for the book AI Engineering (Chip Huyen, 2025)","aie-book 是知名技术作家 Chip Huyen 为其新书《AI Engineering》打造的配套开源资源库。它并非传统的代码教程，而是一套面向实战的 AI 工程化指南，旨在帮助从业者系统性地掌握如何利用基础模型（如大语言模型和多模态模型）解决现实世界的问题。\n\n在 AI 技术快速迭代的今天，开发者常面临“何时微调模型”、“如何评估应用效果”、“怎样抑制幻觉”以及\"RAG 最佳实践”等关键决策难题。aie-book 通过提供书籍章节摘要、学习笔记、丰富的提示词（Prompt）案例、深度行业案例分析以及独特的对话热力图生成工具，为这些问题提供了经过验证的方法论和评估框架。它强调超越具体工具的生命周期，聚焦于构建可靠、高效且安全 AI 系统的核心原则。\n\n这套资源特别适合 AI 工程师、机器学习研究者以及希望将大模型落地到生产环境的技术决策者。无论你是想从零构建 AI 应用，还是希望优化现有系统，aie-book 都能助你理清思路，避开常见陷阱，建立起扎实的 AI 工程化思维。","# AI Engineering book and other resources\n> _This repo will be updated with more resources in the next few weeks._\n\n- [About the book AI Engineering](#about-the-book)\n    - [Table of contents](ToC.md)\n    - [Chapter summaries](chapter-summaries.md)\n    - [Study notes](study-notes.md)\n- [AI engineering resources](resources.md)\n- [Prompt examples](prompt-examples.md)\n- [Case studies](case-studies.md)\n- [Misalignment AI](misalignment.md)\n- [Appendix](appendix.md)\n- Fun tools:\n    \n    - [ChatGPT and Claude conversation heatmap generator](scripts\u002Fai-heatmap.ipynb)\n- And more ...\n\n## About the book\nThe availability of foundation models has transformed AI from a specialized discipline into a powerful development tool everyone can use. This book covers the end-to-end process of adapting foundation models to solve real-world problems, encompassing tried-and-true techniques from other engineering fields and techniques emerging with foundation models.\n\n[\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fchiphuyen_aie-book_readme_605d82ce6a01.png\" width=\"250\">](https:\u002F\u002Famzn.to\u002F49j1cGS)[\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fchiphuyen_aie-book_readme_0ab1057d5438.png\" width=\"250\">](https:\u002F\u002Famzn.to\u002F49j1cGS)\n\nThe book is available on:\n- [Amazon](https:\u002F\u002Famzn.to\u002F49j1cGS)\n- [O'Reilly](https:\u002F\u002Foreillymedia.pxf.io\u002Fc\u002F5719111\u002F2146021\u002F15173)\n- [Kindle](https:\u002F\u002Famzn.to\u002F3Vq2ryu)\n\nand most places where technical books are sold.\n\n_This is NOT a tutorial book, so it doesn't have a lot of code snippets._\n\n## What this book is about\nThis book provides a framework for adapting foundation models, which include both large language models (LLMs) and large multimodal models (LMMs), to specific applications. It not only outlines various solutions for building an AI application but also raises questions you can ask to evaluate the best solution for your needs. Here are just some of the many questions that this book can help you answer:\n\n1. Should I build this AI application?\n1. How do I evaluate my application? Can I use AI to evaluate AI outputs?\n1. What causes hallucinations? How do I detect and mitigate hallucinations?\n1. What are the best practices for prompt engineering?\n1. Why does RAG work? What are the strategies for doing RAG?\n1. What’s an agent? How do I build and evaluate an agent?\n1. When to finetune a model? When not to finetune a model?\n1. How much data do I need? How do I validate the quality of my data?\n1. How do I make my model faster, cheaper, and secure?\n1. How do I create a feedback loop to improve my application continually?\n\nThe book will also help you navigate the overwhelming AI landscape: types of models, evaluation benchmarks, and a seemingly infinite number of use cases and application patterns.\n\nThe content in this book is illustrated using actual case studies, many of which I’ve worked on, backed by ample references and extensively reviewed by experts from a wide range of backgrounds. Even though the book took two years to write, it draws from my experience working with language models and ML systems from the last decade.\n\nLike my previous book, _[Designing Machine Learning Systems (DMLS)](https:\u002F\u002Famzn.to\u002F4fXVZH2)_, this book focuses on the fundamentals of AI engineering instead of any specific tool or API. Tools become outdated quickly, but fundamentals should last longer.\n\n### Reading _AI Engineering_ (AIE) with _Designing Machine Learning Systems_ (DMLS)\nAIE can be a companion to DMLS. DMLS focuses on building applications on top of traditional ML models, which involves more tabular data annotations, feature engineering, and model training. AIE focuses on building applications on top of foundation models, which involves more prompt engineering, context construction, and parameter-efficient finetuning. Both books are self-contained and modular, so you can read either book independently.\n\nSince foundation models are ML models, some concepts are relevant to working with both. If a topic is relevant to AIE but has been discussed extensively in DMLS, it’ll still be covered in this book, but to a lesser extent, with pointers to relevant resources. \n\nNote that many topics are covered in DMLS but not in AIE, and vice versa. The first chapter of this book also covers the differences between traditional ML engineering and AI engineering.\n\nA real-world system often involves both traditional ML models and foundation models, so knowledge about working with both is often necessary.\n\n## Who this book is for\n\nThis book is for anyone who wants to leverage foundation models to solve real-world problems. This is a technical book, so the language of this book is geared towards technical roles, including AI engineers, ML engineers, data scientists, engineering managers, and technical product managers. This book is for you if you can relate to one of the following scenarios:\n* You’re building or optimizing an AI application, whether you’re starting from scratch or looking to move beyond the demo phase into a production-ready stage. You may also be facing issues like hallucinations, security, latency, or costs, and need targeted solutions.\n* You want to streamline your team’s AI development process, making it more systematic, faster, and reliable.\n* You want to understand how your organization can leverage foundation models to improve the business’s bottom line and how to build a team to do so.\n\nYou can also benefit from the book if you belong to one of the following groups:\n* Tool developers who want to identify underserved areas in AI engineering to position your products in the ecosystem.\n* Researchers who want to understand better AI use cases.\n* Job candidates seeking clarity on the skills needed to pursue a career as an AI engineer.\n* Anyone wanting to better understand AI's capabilities and limitations, and how it might affect different roles.\n\nI love getting to the bottom of things, so some sections dive a bit deeper into the technical side. While many early readers like the detail, I know it might not be for everyone. I’ll give you a heads-up before things get too technical. Feel free to skip ahead if it feels a little too in the weeds!\n\n\n## Reviews\n- _\"This book offers a comprehensive, well-structured guide to the essential aspects of building generative AI systems. A must-read for any professional looking to scale AI across the enterprise.\"_ - Vittorio Cretella, former global CIO at P&G and Mars\n\n- _\"Chip Huyen gets generative AI. She is a remarkable teacher and writer whose work has been instrumental in helping teams bring AI into production. Drawing on her deep expertise, AI Engineering is a comprehensive and holistic guide to building generative AI applications in production.\"_ - Luke Metz, co-creator of ChatGPT, ex-research manager @ OpenAI\n\n- _\"Every AI engineer building real-world applications should read this book. It’s a vital guide to end-to-end AI system design, from model development and evaluation to large-scale deployment and operation.\"_ - Andrei Lopatenko, Director Search and AI, Neuron7\n\n- _\"This book serves as an essential guide for building AI products that can scale. Unlike other books that focus on tools or current trends that are constantly changing, Chip delivers timeless foundational knowledge. Whether you're a product manager or an engineer, this book effectively bridges the collaboration gap between cross-functional teams, making it a must-read for anyone involved in AI development.\"_ - Aileen Bui, AI Product Operations Manager, Google\n\n- _\"This is the definitive segue into AI Engineering from one of the greats of ML Engineering! Chip has seen through successful projects and careers at every stage of a company and for the first time ever condensed her expertise for new AI Engineers entering the field.\"_ - swyx, Curator, AI.Engineer\n\n- _\"AI Engineering is a practical guide that provides the most up-to-date information on AI development, making it approachable for novice and expert leaders alike. This book is an essential resource for anyone looking to build robust and scalable AI systems.\"_ - Vicki Reyzelman, Chief AI Solutions Architect, Mave Sparks\n\n- _\"AI Engineering is a comprehensive guide that serves as an essential reference for both understanding and implementing AI systems in practice.\"_ - Han Lee, Director - Data Science, Moody's.\n\n- _\"AI Engineering is an essential guide for anyone building software with Generative AI! It demystifies the technology, highlights the importance of evaluation, and shares what should be done to achieve quality before starting with costly fine-tuning.\"_ - Rafal Kawala, Senior AI Engineering Director, 16 years of experience working in a Fortune 500 company\n\nSee what people are talking about the book on Twitter [@aisysbooks](https:\u002F\u002Ftwitter.com\u002Faisysbooks\u002Flikes)!\n\n## Acknowledgments\nThis book would've taken a lot longer to write and missed many important topics if it wasn't for so many wonderful people who helped me through the process.\n\nBecause the timeline for the project was tight—two years for a 150,000-word book that covers so much ground—I'm grateful to the technical reviewers who put aside their precious time to review this book so quickly.\n\n[Luke Metz](https:\u002F\u002Fx.com\u002Fluke_metz) is an amazing soundboard who checked my assumptions and prevented me from going down the wrong path. [Han-chung Lee](https:\u002F\u002Fwww.linkedin.com\u002Fin\u002Fhanchunglee\u002F), always up to date with the latest AI news and community development, pointed me toward resources that I missed. Luke and Han were the first to review my drafts before I sent them to the next round of technical reviewers, and I'm forever indebted to them for tolerating my follies and mistakes.\n\nHaving led AI innovation at Fortune 500 companies, [Vittorio Cretella](https:\u002F\u002Fwww.linkedin.com\u002Fin\u002Fvittorio-cretella\u002F) and [Andrei Lopatenko](https:\u002F\u002Fwww.linkedin.com\u002Fin\u002Flopatenko\u002F) provided invaluable feedback that combined deep technical expertise with executive insights. [Vicki Reyzelman](https:\u002F\u002Fwww.linkedin.com\u002Fin\u002Fvickireyzelman\u002F) helped me ground my content and keep it relevant for readers with a software engineering background.\n\n[Eugene Yan](https:\u002F\u002Feugeneyan.com\u002F), a dear friend and amazing applied scientist, provided me with technical and emotional support. Shawn Wang ([swyx](https:\u002F\u002Fx.com\u002Fswyx)), provided an important vibe check that helped me feel more confident about the book. [Sanyam Bhutani](https:\u002F\u002Fx.com\u002Fbhutanisanyam1) is one of the best learners and most humble souls I know, who not only gave thoughtful written feedback but also recorded videos to explain his feedback.\n\nKyle Krannen is a star deep learning lead who interviewed his colleagues and shared with me an amazing writeup about their finetuning process, which guided the finetuning chapter. [Mark Saroufim](https:\u002F\u002Fx.com\u002Fmarksaroufim), an inquisitive mind who always has his pulse on the most interesting problems, introduced me to great resources on efficiency. Both Kyle and Mark's feedback was critical in writing Chapters 7 and 9.\n\n[Kittipat \"Bot\" Kampa](https:\u002F\u002Fwww.linkedin.com\u002Fin\u002Fkittipat-bot-kampa-1b1965\u002F), on top of answering my many questions, shared with me a detailed visualization of how he thinks about AI platform. I appreciate [Denys Linkov](https:\u002F\u002Fwww.linkedin.com\u002Fin\u002Fdenyslinkov\u002F)'s systematic approach to evaluation and platform development. [Chetan Tekur](https:\u002F\u002Fwww.linkedin.com\u002Fin\u002Fchetantekur\u002F) gave great examples that helped me structure AI application patterns. I'd also like to thank [Alex (Shengzhi Li) Li](https:\u002F\u002Fwww.linkedin.com\u002Fin\u002Ffindalexli\u002F) and [Hien Luu](https:\u002F\u002Fwww.linkedin.com\u002Fin\u002Fhienluu\u002F) for their thoughtful feedback on my draft on AI architecture.\n\n[Aileen Bui](https:\u002F\u002Fwww.linkedin.com\u002Fin\u002Faileenbui\u002F) is a treasure who shared unique feedback and examples from a product manager's perspective. Thanks [Todor Markov](https:\u002F\u002Fwww.linkedin.com\u002Fin\u002Ftodor-markov-4aa38a67\u002F) for the actionable advice on the RAG and Agents chapter. Thanks [Tal Kachman](https:\u002F\u002Fwww.linkedin.com\u002Fin\u002Ftal-kachman\u002F) for jumping in at the last minute to push the finetuning chapter over the finish line. \n\nThere are so many wonderful people whose company and conversations gave me ideas that guide the content of this book. I tried my best to include the names of everyone who has helped me here, but due to the inherent faultiness of human memory, I undoubtedly neglected to mention many. If I forgot to include your name, please know that it wasn't because I don't appreciate your contribution, and please kindly remind me so that I can rectify as soon as possible!\n\nAndrew Francis, Anish Nag, [Anthony Galczak](https:\u002F\u002Fwww.linkedin.com\u002Fin\u002Fwgalczak\u002F), [Anton Bacaj](https:\u002F\u002Fx.com\u002Fabacaj), Balázs Galambosi, Charles Frye, Charles Packer, Chris Brousseau, Eric Hartford, Goku Mohandas, Hamel Husain, Harpreet Sahota, Hassan El Mghari, Huu Nguyen, Jeremy Howard, Jesse Silver, John Cook, [Juan Pablo Bottaro](https:\u002F\u002Fwww.linkedin.com\u002Fin\u002Fjuan-pablo-bottaro\u002F), Kyle Gallatin, Lance Martin, Lucio Dery, Matt Ross, Maxime Labonne, Miles Brundage, Nathan Lambert, Omar Khattab, [Phong Nguyen](https:\u002F\u002Fwww.linkedin.com\u002Fin\u002Fxphongvn\u002F), Purnendu Mukherjee, Sam Reiswig, Sebastian Raschka, Shahul ES, Sharif Shameem, Soumith Chintala, Teknium, Tim Dettmers, Undi5, Val Andrei Fajardo, Vern Liang, Victor Sanh, Wing Lian, Xiquan Cui, Ying Sheng, and Kristofer.\n\nI'd like to thank all early readers who have also reached out with feedback. Douglas Bailley is a super reader who shared so much thoughtful feedback. Nutan Sahoo for suggesting an elegant way to explain perplexity.\n\nI learned so much from the online discussions with so many. Thanks to everyone who's ever answered my questions, commented on my posts, or sent me an email with your thoughts.\n\nOf course, the book wouldn't have been possible without the team at O'Reilly, especially my development editors (Melissa Potter, Corbin Collins, Jill Leonard) and my production editors (Kristen Brown and Elizabeth Kelly). Liz Wheeler is the most discerning editor I've ever worked with. Nicole Butterfield is a force who oversaw this book from an idea to a final product.\n\nThis book, after all, is an accumulation of invaluable lessons I learned throughout my career. I owe these lessons to my extremely competent and patient coworkers and former coworkers. Every person I've worked with has taught me something new about bringing ML into the world.\n\n---\n\n\u003Cbr>\n\u003Cbr>\n\nChip Huyen, *AI Engineering*. O'Reilly Media, 2025.\n\n    @book{aiebook2025,  \n        address = {USA},  \n        author = {Chip Huyen},  \n        isbn = {978-1801819312},   \n        publisher = {O'Reilly Media},  \n        title = {{AI Engineering}},  \n        year = {2025}  \n    }\n","# AI 工程书籍及其他资源\n> _本仓库将在接下来的几周内持续更新更多资源。_\n\n- [关于《AI 工程》一书](#about-the-book)\n    - [目录](ToC.md)\n    - [各章概要](chapter-summaries.md)\n    - [学习笔记](study-notes.md)\n- [AI 工程资源](resources.md)\n- [提示词示例](prompt-examples.md)\n- [案例研究](case-studies.md)\n- [对齐问题与 AI](misalignment.md)\n- [附录](appendix.md)\n- 有趣工具：\n    \n    - [ChatGPT 和 Claude 对话热力图生成器](scripts\u002Fai-heatmap.ipynb)\n- 还有更多……\n\n## 关于本书\n基础模型的出现，使 AI 从一门专业学科转变为人人都能使用的强大开发工具。本书涵盖了将基础模型应用于解决现实世界问题的端到端流程，既包括来自其他工程领域的成熟技术，也涵盖了随着基础模型兴起而出现的新方法。\n\n[\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fchiphuyen_aie-book_readme_605d82ce6a01.png\" width=\"250\">](https:\u002F\u002Famzn.to\u002F49j1cGS)[\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fchiphuyen_aie-book_readme_0ab1057d5438.png\" width=\"250\">](https:\u002F\u002Famzn.to\u002F49j1cGS)\n\n本书可在以下平台购买：\n- [亚马逊](https:\u002F\u002Famzn.to\u002F49j1cGS)\n- [O'Reilly](https:\u002F\u002Foreillymedia.pxf.io\u002Fc\u002F5719111\u002F2146021\u002F15173)\n- [Kindle](https:\u002F\u002Famzn.to\u002F3Vq2ryu)\n\n以及大多数销售技术类书籍的渠道。\n\n_本书并非教程类书籍，因此不包含大量代码片段。_\n\n## 本书内容概述\n本书提供了一套框架，用于将包括大型语言模型（LLM）和大型多模态模型（LMM）在内的基础模型适配到具体应用场景中。它不仅概述了构建 AI 应用的各种方案，还提出了可以帮助您评估最适合自身需求的解决方案的问题。以下是本书能够帮助您解答的部分问题：\n\n1. 我是否应该构建这个 AI 应用？\n1. 如何评估我的应用？能否使用 AI 来评估 AI 的输出？\n1. 幻觉现象由什么引起？如何检测并缓解幻觉？\n1. 提示工程的最佳实践是什么？\n1. RAG 为何有效？实施 RAG 的策略有哪些？\n1. 什么是智能体？如何构建并评估一个智能体？\n1. 何时需要对模型进行微调？何时不需要？\n1. 我需要多少数据？如何验证数据的质量？\n1. 如何让我的模型更快、更便宜且更安全？\n1. 如何建立反馈循环以持续改进我的应用？\n\n此外，本书还将帮助您理清纷繁复杂的 AI 领域：模型类型、评估基准，以及数量看似无穷无尽的应用场景和模式。\n\n书中内容通过实际案例加以说明，其中许多是我亲自参与过的项目，并辅以丰富的参考文献，同时经过来自不同背景的专家广泛审阅。尽管本书耗时两年完成，但它基于我在过去十年中使用语言模型和机器学习系统的工作经验。\n\n与我之前的著作《设计机器学习系统》（DMLS）一样，本书聚焦于 AI 工程的基础知识，而非特定的工具或 API。工具会迅速过时，而基础原理则更为持久。\n\n### 结合阅读《AI 工程》（AIE）与《设计机器学习系统》（DMLS）\nAIE 可以作为 DMLS 的补充读物。DMLS 主要关注在传统机器学习模型之上构建应用，涉及更多的表格型数据标注、特征工程和模型训练；而 AIE 则侧重于在基础模型之上构建应用，涉及更多的提示工程、上下文构建和参数高效的微调。两本书均为独立且模块化的结构，您可以单独阅读其中任何一本。\n\n由于基础模型本质上也是机器学习模型，因此部分概念在两者中都适用。如果某个主题虽然与 AIE 相关，但在 DMLS 中已有详尽讨论，则本书仍会提及该主题，但篇幅会相对较少，并提供相关资源的链接。\n\n需要注意的是，许多主题仅在 DMLS 或 AIE 中有所涉及，反之亦然。本书的第一章还详细探讨了传统机器学习工程与 AI 工程之间的区别。\n\n在实际系统中，往往同时涉及传统机器学习模型和基础模型，因此掌握两者的相关知识通常是必要的。\n\n## 本书适合哪些读者\n\n本书面向所有希望利用基础模型解决现实世界问题的人士。这是一本技术类书籍，因此其语言风格主要针对技术岗位，包括 AI 工程师、机器学习工程师、数据科学家、工程经理和技术产品经理等。如果您符合以下任一情况，本书都将对您有所帮助：\n* 您正在构建或优化 AI 应用，无论是从零开始，还是希望从演示阶段进入生产就绪状态。您可能还面临幻觉、安全性、延迟或成本等问题，亟需针对性的解决方案。\n* 您希望简化团队的 AI 开发流程，使其更加系统化、高效化和可靠化。\n* 您希望了解组织如何利用基础模型提升业务效益，以及如何组建相应的团队来实现这一目标。\n\n此外，以下群体也能从本书中受益：\n* 工具开发者：希望识别 AI 工程领域中尚未被充分覆盖的空白点，从而更好地定位您的产品在生态系统中的位置。\n* 研究人员：希望更深入地理解 AI 的各种应用场景。\n* 求职者：希望明确从事 AI 工程师职业所需的技能。\n* 任何希望更好地理解 AI 的能力与局限性，以及它可能对不同角色产生的影响的人士。\n\n我喜欢深入探究事物的本质，因此书中某些章节会更深入地剖析技术细节。尽管许多早期读者喜欢这种细致入微的描述，但我明白这并不一定适合所有人。当内容变得过于技术化时，我会提前告知您，请随时跳过那些感觉过于复杂的部分！\n\n## 书评\n- _“本书为构建生成式AI系统的关键环节提供了一套全面且结构清晰的指南。对于任何希望在企业范围内规模化应用AI的专业人士而言，这都是一本必读之作。”_ —— 维托里奥·克雷特拉，宝洁与玛氏前全球首席信息官\n\n- _“希普·休恩深谙生成式AI之道。她是一位杰出的导师与作家，其著作对帮助团队将AI落地生产发挥了至关重要的作用。凭借其深厚的专业积累，《AI工程》是一部全面而系统性的指南，指导如何在生产环境中构建生成式AI应用。”_ —— 卢克·梅茨，ChatGPT联合开发者、前OpenAI研究经理\n\n- _“每一位构建真实世界应用的AI工程师都应当阅读此书。它是一本关于端到端AI系统设计的重要指南，涵盖从模型开发与评估，到大规模部署与运维的全流程。”_ —— 安德烈·洛帕滕科，Neuron7搜索与AI总监\n\n- _“本书是构建可规模化AI产品的必备指南。不同于那些聚焦于工具或瞬息万变的最新趋势的书籍，希普提供了历久弥新的基础性知识。无论你是产品经理还是工程师，这本书都能有效弥合跨职能团队之间的协作鸿沟，因此对于所有参与AI开发的人来说，都堪称必读之作。”_ —— 艾琳·布依，谷歌AI产品运营经理\n\n- _“这是机器学习工程领域大师向AI工程领域过渡的权威之作！希普在其职业生涯中亲历了公司各个发展阶段的成功项目，并首次将她的丰富经验浓缩成册，献给刚入行的新一代AI工程师。”_ —— swyx，AI.Engineer策展人\n\n- _“《AI工程》是一本实用指南，提供了最前沿的AI开发资讯，既适合初学者，也便于资深领导者理解。对于任何希望构建稳健且可扩展的AI系统的人来说，这本书都是不可或缺的资源。”_ —— 维姬·雷兹尔曼，Mave Sparks首席AI解决方案架构师\n\n- _“《AI工程》是一部内容全面的指南，既是理解AI系统的关键参考，也是将其付诸实践的实用手册。”_ —— 韩·李，穆迪公司数据科学总监\n\n- _“《AI工程》是所有使用生成式AI进行软件开发人员的必备指南！它揭开了这项技术的神秘面纱，强调了评估的重要性，并分享了在开始代价高昂的微调之前，应当采取哪些措施来确保质量。”_ —— 拉法尔·卡瓦拉，高级AI工程总监，拥有16年财富500强企业工作经验\n\n快来Twitter上看看大家对这本书的讨论吧[@aisysbooks](https:\u002F\u002Ftwitter.com\u002Faisysbooks\u002Flikes)！\n\n## 致谢\n如果没有众多杰出人士在写作过程中给予的帮助，这本书的完成将会耗费更长的时间，并且会遗漏许多重要内容。\n\n由于项目时间紧迫——仅用两年时间就要完成一部涵盖广泛内容、字数达15万字的书籍——我非常感激那些技术审稿人，他们不惜牺牲宝贵时间，迅速完成了本书的审阅工作。\n\n[Luke Metz](https:\u002F\u002Fx.com\u002Fluke_metz) 是一位极好的“思想伙伴”，他帮助我检验假设，避免我走入误区。[Han-chung Lee](https:\u002F\u002Fwww.linkedin.com\u002Fin\u002Fhanchunglee\u002F) 始终紧跟人工智能领域的最新动态和社区发展，为我指出了许多被我忽略的重要资源。Luke 和 Han 是在我将稿件递交给下一轮技术审稿人之前，最早审阅我的初稿的人，我永远感激他们对我种种失误与不足的包容。\n\n曾在世界500强企业领导人工智能创新的 [Vittorio Cretella](https:\u002F\u002Fwww.linkedin.com\u002Fin\u002Fvittorio-cretella\u002F) 和 [Andrei Lopatenko](https:\u002F\u002Fwww.linkedin.com\u002Fin\u002Flopatenko\u002F) 提供了极具价值的反馈，他们的意见既深入技术细节，又兼具高管视角。[Vicki Reyzelman](https:\u002F\u002Fwww.linkedin.com\u002Fin\u002Fvickireyzelman\u002F) 则帮助我确保内容扎实可靠，并使其对具备软件工程背景的读者更具相关性。\n\n[Eugene Yan](https:\u002F\u002Feugeneyan.com\u002F) 是一位挚友兼杰出的应用科学家，他不仅在技术上给予了我支持，也在情感上给了我莫大的鼓励。Shawn Wang（[swyx](https:\u002F\u002Fx.com\u002Fswyx)）则为我提供了重要的方向性建议，让我对整本书的信心倍增。[Sanyam Bhutani](https:\u002F\u002Fx.com\u002Fbhutanisanyam1) 是我所认识的最佳学习者之一，也是一位极为谦逊的人。他不仅撰写了细致入微的文字反馈，还录制视频来进一步阐释自己的意见。\n\nKyle Krannen 是一位卓越的深度学习负责人，他采访了自己的同事，并与我分享了一份关于他们微调流程的精彩报告，这为微调章节的撰写提供了重要指导。[Mark Saroufim](https:\u002F\u002Fx.com\u002Fmarksaroufim)，这位充满求知欲、时刻关注最前沿问题的人，向我推荐了许多关于效率优化的优质资源。Kyle 和 Mark 的反馈对于第7章和第9章的写作至关重要。\n\n[Kittipat “Bot” Kampa](https:\u002F\u002Fwww.linkedin.com\u002Fin\u002Fkittipat-bot-kampa-1b1965\u002F) 除了耐心解答我提出的诸多问题外，还与我分享了他对人工智能平台的独特思考方式及其详细可视化呈现。我同样感谢 [Denys Linkov](https:\u002F\u002Fwww.linkedin.com\u002Fin\u002Fdenyslinkov\u002F) 在评估与平台开发方面所展现出的系统化方法。[Chetan Tekur](https:\u002F\u002Fwww.linkedin.com\u002Fin\u002Fchetantekur\u002F) 则提供了许多精彩的案例，帮助我梳理了人工智能应用模式。此外，我还想感谢 [Alex (Shengzhi Li) Li](https:\u002F\u002Fwww.linkedin.com\u002Fin\u002Ffindalexli\u002F) 和 [Hien Luu](https:\u002F\u002Fwww.linkedin.com\u002Fin\u002Fhienluu\u002F) 对我关于人工智能架构的草稿所提出的深刻见解。\n\n[Aileen Bui](https:\u002F\u002Fwww.linkedin.com\u002Fin\u002Faileenbui\u002F) 是一位难得的宝藏，她从产品经理的角度出发，提供了独到的反馈和实例。感谢 [Todor Markov](https:\u002F\u002Fwww.linkedin.com\u002Fin\u002Ftodor-markov-4aa38a67\u002F) 对 RAG 和 Agent 章节提出的切实可行的建议。同时也要感谢 [Tal Kachman](https:\u002F\u002Fwww.linkedin.com\u002Fin\u002Ftal-kachman\u002F) 在最后关头挺身而出，推动微调章节顺利完成。\n\n还有许许多多令人敬佩的朋友，他们的陪伴与交流为本书的内容提供了源源不断的灵感。我已尽力在此列出所有曾给予我帮助的人士姓名，但由于人类记忆难免有疏漏，我肯定遗漏了不少名字。如果我的名单中遗漏了您的大名，请您谅解，这绝非我不重视您的贡献；请您不吝提醒我，我会尽快予以补正！\n\nAndrew Francis、Anish Nag、[Anthony Galczak](https:\u002F\u002Fwww.linkedin.com\u002Fin\u002Fwgalczak\u002F)、[Anton Bacaj](https:\u002F\u002Fx.com\u002Fabacaj)、Balázs Galambosi、Charles Frye、Charles Packer、Chris Brousseau、Eric Hartford、Goku Mohandas、Hamel Husain、Harpreet Sahota、Hassan El Mghari、Huu Nguyen、Jeremy Howard、Jesse Silver、John Cook、[Juan Pablo Bottaro](https:\u002F\u002Fwww.linkedin.com\u002Fin\u002Fjuan-pablo-bottaro\u002F)、Kyle Gallatin、Lance Martin、Lucio Dery、Matt Ross、Maxime Labonne、Miles Brundage、Nathan Lambert、Omar Khattab、[Phong Nguyen](https:\u002F\u002Fwww.linkedin.com\u002Fin\u002Fxphongvn\u002F)、Purnendu Mukherjee、Sam Reiswig、Sebastian Raschka、Shahul ES、Sharif Shameem、Soumith Chintala、Teknium、Tim Dettmers、Undi5、Val Andrei Fajardo、Vern Liang、Victor Sanh、Wing Lian、Xiquan Cui、Ying Sheng，以及 Kristofer。\n\n我也要感谢所有早期读者，他们纷纷主动提出宝贵意见。Douglas Bailley 是一位超级热心的读者，他分享了大量富有洞见的反馈。Nutan Sahoo 则提出了一个优雅的方式来解释困惑度这一概念。\n\n在与各位在线交流的过程中，我也受益匪浅。感谢每一位曾经回答过我问题、评论过我的帖子，或通过邮件与我分享想法的朋友。\n\n当然，没有 O'Reilly 团队的支持，尤其是我的编辑团队（Melissa Potter、Corbin Collins、Jill Leonard）以及制作编辑团队（Kristen Brown 和 Elizabeth Kelly），这本书根本无法问世。Liz Wheeler 是我合作过的最具洞察力的编辑。Nicole Butterfield 则是整个项目的幕后推手，她全程主导了本书从构思到最终成品的每一个环节。\n\n归根结底，这本书是我职业生涯中所积累的宝贵经验的结晶。这些经验离不开我那些能力出众且耐心十足的同事及前同事们。与我共事过的每一个人，都教会了我如何将机器学习真正落地并造福社会。\n\n---\n\n\u003Cbr>\n\u003Cbr>\n\nChip Huyen，《AI 工程》。O'Reilly Media 出版社，2025年。\n\n    @book{aiebook2025,  \n        address = {美国},  \n        author = {Chip Huyen},  \n        isbn = {978-1801819312},   \n        publisher = {O'Reilly Media},  \n        title = {{AI 工程}},  \n        year = {2025}  \n    }","# aie-book 快速上手指南\n\n`aie-book` 并非一个需要安装运行的软件库或框架，而是 Chip Huyen 所著《AI Engineering》一书的开源配套资源仓库。它提供了书籍目录、章节摘要、学习笔记、提示词（Prompt）示例、案例研究以及实用的辅助脚本。开发者主要通过阅读文档和运行提供的示例脚本来获取知识。\n\n## 环境准备\n\n由于本仓库主要包含 Markdown 文档、Jupyter Notebook 脚本及相关资源链接，对环境要求极低。\n\n*   **操作系统**：Windows, macOS, 或 Linux 均可。\n*   **前置依赖**：\n    *   **Git**：用于克隆仓库。\n    *   **浏览器**：用于直接在线阅读 GitHub 上的 Markdown 文件。\n    *   **Python (可选)**：仅当你需要运行仓库中的 `scripts\u002Fai-heatmap.ipynb` 等交互式脚本时需要安装。建议版本 Python 3.8+。\n    *   **Jupyter Lab \u002F Notebook (可选)**：用于运行和编辑 `.ipynb` 脚本文件。\n\n## 安装步骤\n\n你无需通过包管理器（如 pip 或 npm）安装此工具，只需将代码仓库克隆到本地即可。\n\n1.  **克隆仓库**\n    打开终端或命令行工具，执行以下命令：\n    ```bash\n    git clone https:\u002F\u002Fgithub.com\u002Fchiphuyen\u002Faie-book.git\n    ```\n\n    *国内加速方案*：如果访问 GitHub 速度较慢，可以使用国内镜像源（如 Gitee 镜像，若有）或通过代理加速。若无特定镜像，可尝试使用 `gitclone.com` 服务：\n    ```bash\n    git clone https:\u002F\u002Fgitclone.com\u002Fgithub.com\u002Fchiphuyen\u002Faie-book.git\n    ```\n\n2.  **进入目录**\n    ```bash\n    cd aie-book\n    ```\n\n3.  **安装 Python 依赖（仅针对运行脚本）**\n    如果你打算运行 `scripts` 目录下的 Notebook 脚本（如热力图生成器），请安装基础数据科学栈：\n    ```bash\n    pip install jupyter pandas matplotlib seaborn\n    ```\n    *注：建议使用国内镜像源加速安装*\n    ```bash\n    pip install -i https:\u002F\u002Fpypi.tuna.tsinghua.edu.cn\u002Fsimple jupyter pandas matplotlib seaborn\n    ```\n\n## 基本使用\n\n### 1. 在线阅读与学习（推荐）\n最直接的使用方式是直接在 GitHub 或本地编辑器中阅读核心资源文件：\n\n*   **查看目录结构**：打开 `ToC.md` 了解全书架构。\n*   **学习核心概念**：阅读 `chapter-summaries.md` 获取章节精华总结。\n*   **获取 Prompt 灵感**：查阅 `prompt-examples.md` 获取实际的提示词工程案例。\n*   **研究真实案例**：参考 `case-studies.md` 了解端到端的 AI 应用构建过程。\n\n### 2. 运行趣味工具示例\n仓库提供了一个生成 ChatGPT 和 Claude 对话热力图的脚本。以下是运行该示例的步骤：\n\n1.  启动 Jupyter Notebook：\n    ```bash\n    jupyter notebook scripts\u002Fai-heatmap.ipynb\n    ```\n2.  在浏览器打开的界面中，依次运行单元格（Cell）。\n    *   该脚本通常用于可视化对话数据中的模式（需确保你有相应的对话数据文件或按照脚本内的说明生成模拟数据）。\n    *   脚本将输出对话互动的热力图，帮助分析模型行为。\n\n### 3. 获取实体书资源\n本仓库是书籍《AI Engineering》的补充材料。若需深入学习系统性知识，建议结合实体书阅读：\n*   **购买渠道**：可通过 [Amazon](https:\u002F\u002Famzn.to\u002F49j1cGS) 或 [O'Reilly](https:\u002F\u002Foreillymedia.pxf.io\u002Fc\u002F5719111\u002F2146021\u002F15173) 获取。\n*   **核心理念**：本书侧重于基础模型（LLMs\u002FLMMs）的工程化落地，涵盖评估、幻觉处理、RAG 策略、Agent 构建及微调时机等关键问题，而非单纯的代码教程。","某初创团队正试图将大语言模型集成到客服系统中，但在技术选型和效果评估阶段陷入了盲目试错的困境。\n\n### 没有 aie-book 时\n- 团队在“微调模型”还是“构建 RAG 系统”之间犹豫不决，缺乏系统的决策框架，导致资源浪费在错误的技术路线上。\n- 面对模型产生的幻觉问题，开发人员只能凭直觉修补提示词，无法定位根本原因或缺乏科学的检测与缓解策略。\n- 由于缺乏端到端的工程视角，团队不知道如何建立有效的反馈闭环，导致应用上线后迭代缓慢，难以持续提升准确率。\n- 成员被市面上层出不穷的新工具和新基准测试搞得晕头转向，难以分辨哪些是营销噱头，哪些是真正适用的最佳实践。\n\n### 使用 aie-book 后\n- 借助书中提供的评估框架，团队快速明确了场景需求，果断选择了更适合动态知识库的 RAG 方案，避免了不必要的微调成本。\n- 利用书中关于幻觉成因的深度解析，团队实施了针对性的检测机制和上下文优化策略，显著降低了错误回答率。\n- 参考书中的持续改进章节，团队设计了一套自动化的用户反馈收集与评估流程，使模型表现能够随数据积累不断进化。\n- 通过阅读真实的案例研究和基础原理，团队成员统一了认知，不再盲目追逐新工具，而是专注于构建稳健的工程架构。\n\naie-book 不仅提供了应对具体技术难题的实操指南，更赋予了工程师在复杂 AI  landscape 中做出正确战略决策的系统性思维。","https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fchiphuyen_aie-book_605d82ce.png","chiphuyen","Chip Huyen","https:\u002F\u002Foss.gittoolsai.com\u002Favatars\u002Fchiphuyen_9fc913f3.jpg","AI x stuff",null,"San Francisco","chipro","https:\u002F\u002Fhuyenchip.com","https:\u002F\u002Fgithub.com\u002Fchiphuyen",[83],{"name":84,"color":85,"percentage":86},"Jupyter Notebook","#DA5B0B",100,14757,2134,"2026-04-20T00:34:36","","未说明",{"notes":93,"python":91,"dependencies":94},"该项目主要是一本关于 AI 工程的书籍及其相关资源（如目录、案例研究、提示词示例等），并非一个需要特定运行环境的可执行软件工具或代码库。README 中提到的唯一代码文件是用于生成聊天热图的脚本 (scripts\u002Fai-heatmap.ipynb)，但未在提供的文本中列出具体的环境依赖、操作系统要求或硬件配置。",[],[36,14,13],"2026-03-27T02:49:30.150509","2026-04-20T19:42:11.654962",[],[]]