Hugging Face just open-sourced the most complete AI Agents course on the internet. Here's exactly what you learn: → Unit 1: What agents actually are - LLMs, model families, special tokens → Unit 2: Build with smolagents, LlamaIndex, and LangGraph → Unit 2 Bonus: Trace and evaluate your agents (observability) → Unit 3: Agentic RAG for real-world use cases → Unit 4: Final project - build, test, and get certified on a live leaderboard → Bonus: AI agents playing Pokémon (yes, really) Only need basic Python and LLM knowledge to start. This is the stack serious builders are learning while everyone else is still watching YouTube tutorials. 100% Opensource. Apache 2.0.