Swyx (Shawn Wang)
AI EngineeringSwyx's AI engineering playbook — creator of Latent Space podcast, coined 'AI Engineer' as a role. Curated essays on LLM ops, agent frameworks, and the AI engineer stack.
About this vault
Shawn Wang (Swyx) created the Latent Space podcast and helped define 'AI Engineer' as a distinct role separate from ML researcher. This vault curates his best writing and podcast highlights: the AI engineering stack, agent frameworks, LLM ops patterns, the 'world's fair' thesis on AI startups, and why the AI engineer is the new full-stack developer. From swyx.io and Latent Space. Auto-synced — new content appears here automatically.
8 articles
All 8 articles
Why AI Engineering Matters
AI Engineering 不是新瓶装旧酒——它是一个全新的工程学科，核心能力在于 prompt engineering、evals 和 retrieval，而不是 ML 训练。Swyx 提出 AI Engineer 是 software engineer 和 ML engineer 之间的新角色。
The Rise of Agentic Coding
Agent 编程正在重塑开发流程：从人写代码到人审代码。关键洞察是 coding agent 的价值不在替代程序员，在于把 10x 工程师的工作方式民主化。
2024 AI Year in Review
全年 AI 发展的结构化复盘：模型能力跃迁、开源追赶闭源、multimodal 成标配、agent 从概念到产品。最有价值的是对每个趋势的'so what'分析。
RAG Is Not Dead
Context window 变大不等于 RAG 过时。RAG 的核心价值是可控性和可追溯性——你知道模型在用哪些信息做决策。长 context 是补充不是替代。
Building AI Products in Public
AI 产品的 build in public 策略与传统 SaaS 不同：核心是分享你的 evals 结果和 prompt iteration 过程。透明度建立信任，过程本身就是内容。
The Small Model Revolution
小模型正在吃掉大模型的市场——不是性能追平，而是性价比碾压。关键insight：90%的生产场景不需要 GPT-4 级别，fine-tuned 小模型成本低10-100倍。
AI Engineer World's Fair Recap
AI Engineer 社区从0到1万人的增长复盘。最大收获：开发者社区的增长飞轮是'教你做 > 帮你做 > 一起做'，内容驱动但目标是行动。
The State of Prompt Engineering
Prompt engineering 从黑魔法走向工程学科：结构化 prompting、chain-of-thought、few-shot 的系统化方法论。核心是把不可复现的'感觉'变成可复现的'流程'。
Start reading, not hoarding.
Import this vault to Burn 451 and actually read what matters.
Content attributed to original authors. Burn 451 curates publicly available writing as a reading index. For removal requests, contact @hawking520.