ai-engineering-from-scratch: Building AI Engineering by Hand: 511 Lessons Across 20 Phases

While the adoption rate of AI tools is high, only 18% of developers feel they can use them professionally. This curriculum is a free open-source educational course designed to bridge that gap. It consists of 20 phases and 511 lessons, providing approximately 329 hours of learning material.

Rather than stopping at simple conceptual explanations, the core focus is on directly creating reusable artifacts such as prompts, skills, agents, and MCP servers in each lesson. It supports four languages—Python, TypeScript, Rust, and Julia—and encourages implementing all algorithms, from linear algebra to autonomous swarms, from raw mathematics.

Unlike scattered papers or short video tutorials, it builds up sequentially so that you can understand internal mechanisms, from backpropagation to attention mechanisms, before using PyTorch. You can also integrate with coding agents to create personalized learning plans and check your understanding with step-by-step quizzes.