AI EngineeringMy Learning Notes

Building Applications with Foundation Models: my personal notes and takeaways from reading Chip Huyen's book.

Foundations First

Understand what foundation models are, how they're trained, and why they behave the way they do, including hallucinations.

Foundation models

📊

Evaluation Done Right

Two full chapters on the hardest problem in AI engineering: measuring whether your system actually works.

How to evaluate

✍️

Prompting, RAG & Agents

Get more from a model without retraining it: better instructions, better context, and tool-using agents.

RAG and agents

🎛️

Finetuning & Data

When (and when not) to finetune, plus how to acquire, synthesize, and clean the data that makes it work.

Finetuning

⚡

Faster & Cheaper Inference

Quantization, distillation, batching, and caching to make your model cheap and fast enough for production.

Inference optimization

🏗️

Architecture & Feedback

Put it all together into a real system and build feedback loops that keep improving it over time.

Full architecture

What is this?

These are my personal learning notes from reading the book AI Engineering: Building Applications with Foundation Models by Chip Huyen (O'Reilly, 2025).

As I worked through the book's ~150,000 words across 10 chapters, I distilled each chapter into clear, jargon-light notes in my own words, so I can revisit the key ideas, trade-offs, and decision frameworks quickly. I'm sharing them here in case they help you too.

Who might find this useful?

Anyone learning to build applications on top of foundation models: AI engineers, ML engineers, software engineers, data scientists, engineering managers, and technical product managers, or anyone who just wants to understand how these models work.

These are my notes, not a replacement for the book

This is my own summary of what I took away, so it reflects my understanding and may simplify or skip things. The book contains the case studies, references, diagrams, and depth that make these ideas stick. If you find these notes useful, please buy the book.

How to read it

New to the field? Read The Big Picture, then go chapter by chapter.
Building something now? Jump straight to the chapter that matches your problem.
Need a quick refresher? Use the Cheat Sheet and Glossary.