Free 'AI Engineering from Scratch' repo: 230+ lessons from linear algebra to agent swarms
thisdudelikesAI · x · 2026-08-19
A free GitHub repo called AI Engineering from Scratch offers a 230+ lesson, 20-phase curriculum that its author claims beats most $2,000 bootcamps:
- Agent Engineering phase: agent loop from first principles, planning with TodoWrite and DAGs, short/long-term/episodic memory, context compression, subagents and delegation, PreToolUse/PostToolUse hooks, self-healing, eval-driven development — all with runnable TypeScript and Python code
- LLMs from Scratch phase: build a tokenizer in Python and Rust, pre-train a 124M-param mini GPT, distributed training with FSDP/DeepSpeed, RLHF reward model + PPO, quantization with INT8/GPTQ/AWQ/GGUF
- Later phases cover building a Transformer from scratch, multi-agent swarms (consensus, hierarchical, DAG orchestration) and 5 capstone projects
Every lesson follows a fixed structure — theory, build from scratch, use real frameworks, then ship a reusable artifact (prompt templates, SKILL .md files, MCP servers). Free, in Python, TypeScript, Rust and Julia.
Related event: Free Open-Source AI Engineering Course Spans 230 Lessons(2 posts)→
More from coding & agent
- Sub8 Bot built with Grok 4.6 releases 18 updates in 24h — Daniel_Farinax · 2026-08-19
- Integrating local skills and CLIs with cloud agents — iannuttall · 2026-08-19
- Self-improving agents hit the Ship of Theseus problem: drift without errors — arpit_bhayani · 2026-08-19
- Gortex Engine Indexes Code into Graphs to Cut Token Usage by 50x — tom_doerr · 2026-08-19
- Remote Approvals Fixed My Coding-Agent Bottleneck: Hybrid Mode Over Full Auto — Sharp-Daikon7631 · 2026-08-19
- PDF Extraction Wrecks RAG: Converting to Markdown Cuts Token Costs by 40-65% — Training_Anybody5754 · 2026-08-19