260+ lessons across 20 phases. Build neural networks, transformers, and LLMs from first principles. Python, TypeScript, Rust, Julia.
Implement every algorithm from raw math. No magic wrappers. You write the backprop, the tokenizer, the attention mechanism.
A structured path from calculus and linear algebra through LLMs, agents, multi-agent swarms, and production deployment.
Python, TypeScript, Rust, Julia. Every lesson has runnable code, not slides. Build a GPT, a RAG system, an agent team.
The entire curriculum is on GitHub. Clone it, fork it, learn at your own pace. No paywall, no signup, no gatekeeping.
Click any phase to see its lessons
One command. Everything you need is in the repository.
Start at Phase 0 or jump to where your knowledge begins.
Each lesson explains the concept from first principles.
Implement the algorithm. Run the code. Break things.
Each phase has exercises. No multiple choice -- real code challenges.
Phase 19 ties it all together. Build a production AI system.
Confused by jargon? We built a glossary that tells you what people say vs what things actually mean.
One command to begin your AI engineering journey.
git clone https://github.com/rohitg00/ai-engineering-from-scratch.git