Book
AI Engineering
Chip Huyen
The book I wish existed when I started shipping LLM systems. Covers evals, RAG, fine-tuning, and inference economics with the rigor missing from most blog posts.
Curated, not comprehensive
Resources
A short, opinionated reading list of books, blogs, and YouTube channels that have shaped how I work. Continuously updated.
Books
Long-form reading that has shaped how I think about systems and ML.
- Book
Designing Data-Intensive Applications
Martin Kleppmann
Not an AI book, but the systems book every AI engineer eventually needs. Helped me think about the data and infrastructure layer underneath the models I work with.
- Book
Designing Machine Learning Systems
Chip Huyen
The classical ML systems counterpart to AI Engineering, covering data pipelines, monitoring, feedback loops, and training-serving skew. Still the framing I use for the non-LLM ML work in our portfolio.
Blogs
Engineering writing I keep returning to in production work.
- Blog
Anthropic Engineering
Anthropic
The blog I read most. Building Effective Agents, Effective Context Engineering, and Code Execution with MCP are the three posts I've referenced most often in production work this year.
- Blog
Cursor
Cursor
Honest engineering writing from a team building one of the most-used AI-native products. The posts on context handling and inference cost optimization are sharper than most blog content from larger labs.
- Blog
Google Research
Google Research
Where research papers get translated into readable engineering posts. Useful for staying current on the foundational work that shows up in production systems a year or two later.
- Blog
LangChain
LangChain
Worth following because LangGraph is part of my daily stack. Pattern write-ups, integration guides, and the occasional postmortem on what's working in production agent systems.
YouTube Channels
Channels where the talks and primers are consistently worth the time.
- YouTube
AI Engineer
AI Engineer Summit
Conference talks from practitioners actually shipping LLM systems in production. Where I find most of the talks worth watching, including Mahesh Murag's MCP workshop, Hamel Husain on evals, and Karpathy keynotes.
- YouTube
IBM Technology
IBM
Short, well-explained primers on the systems and concepts underneath modern AI: vector databases, RAG, MCP, and agent architectures. Good for filling gaps fast without sitting through a two-hour conference talk.
- YouTube
Y Combinator
Y Combinator
The business and product side of AI that pure engineering channels miss. Founder interviews, AI Startup School keynotes, and the ongoing conversation about what's actually getting built and sold.