ai-engineering-from-scratch

Why we included this project

Most AI courses teach you to glue libraries together; this one walks you through building the algorithms themselves. It's a free curriculum of 503 lessons across 20 phases, starting with linear algebra and moving through backpropagation, tokenizers, attention, and transformers into LLM engineering, then on to agents, multi-agent swarms, and production infrastructure. Each lesson follows a build-it-then-use-it pattern: you implement the algorithm from raw math first, then run the same thing through PyTorch or sklearn, so you see what the framework is doing under the hood. The material spans Python, TypeScript, Rust, and Julia, and every lesson ships a reusable artifact like a prompt, skill, agent, or MCP server. You can read lessons on the web, clone the repo, or install the curriculum as a skill into Claude Code, Cursor, or Codex, where your coding agent acts as a tutor with a placement quiz and a personalized study plan.