GitHub radar
Free course: learn how AI models, search, and agents actually work
A free, open-source, 542-lesson course: you write the code for models, retrieval, and agents yourself, test it, and see why it breaks.
A course where you build pieces of an AI system yourself — from math to production — and prove it works with code, not just reading.
When it helps
Useful if you already work with AI agents and want to understand what happens under the hood, not just write prompts. Don't try to finish it all — pick one path instead of all 542 lessons.
Pros
- Free and open source (MIT license)
- Three projects with automated checks, not just explanations
- Built-in quiz picks the right starting section for you
- Runs on a regular laptop, no GPU needed
Cons
- 542 lessons across 20 phases — finishing it all takes hundreds of hours
- Some lessons need Node.js and Rust, not just Python
How to set it up — step by step
- 1Open your coding agent (Claude Code or Codex) inside your project folder.
- 2Send the agent the text from the block below.
- 3Take the built-in 10-question quiz — it picks your starting phase.
- 4Check the LEARNING.md file it saves — your personal study plan.
- 5Pick one path — agents, MCP, or skills — instead of all 542 lessons.
- 6Try one of the three projects with automated graders, like Retrieval Evaluation Lab.
Text for your agent
Copy this and send it to your agent — Claude Code, Codex, any of them:
Install the AI Engineering From Scratch course as a Claude Code skill (`npx skills add rohitg00/ai-engineering-from-scratch`), run the built-in placement quiz to figure out which phase I should start at, and begin the first lesson via /start-learning.
Runs on a regular laptop — no GPU needed; most lessons only need Python, a few also need Node.js or Rust.
Open on GitHub▌ More finds