AI Engineering from Scratch is a free, open source curriculum that teaches AI engineering by building every core algorithm by hand before any framework is introduced. It spans 20 phases, from linear algebra through tokenizers, backpropagation, attention and agent loops to autonomous multi agent systems, with lessons in Python, TypeScript, Rust and Julia.
It is intended for developers and students who want to understand what libraries such as PyTorch compute rather than only call them. Each lesson follows the same loop: read the problem, derive the math, write the code, run the test and keep the resulting artifact, which may be a prompt, a skill, an agent or an MCP server. Learning paths group the material into domains covering building and deploying AI applications, software engineering fundamentals, agent assisted engineering, and product judgment and delivery. The course can also be taught inside a coding agent: installing its skill package into Claude, Cursor, Codex or any agent that reads SKILL.md files turns the agent into a tutor that runs a placement quiz, builds a personalized path and teaches lessons interactively in the terminal.
There is no charge. The curriculum is released under the MIT license, maintained by Rohit Ghumare and contributors, and runs on the learner's own machine.
Lessons are authored with AI assistance and reviewed by a human against primary sources such as RFCs, specifications and research papers, with corrections tracked publicly on GitHub.
Related Tools
Other development & code tools in the same category
