The Hugging Face Agents Course
A free course that teaches how to build AI agents, moving from the basics of agents and LLMs through frameworks such as smolagents, LlamaIndex and LangGraph to a final, automatically evaluated project.
The course is organized into numbered units: an introduction to agents, LLMs and the underlying model family tree; a bonus unit on fine-tuning an LLM for function-calling; an overview of agent frameworks followed by dedicated units on the smolagents, LlamaIndex and LangGraph frameworks; a bonus unit on observability and evaluation; a unit on agentic retrieval-augmented generation across different frameworks; and a final project in which the student's own agent is automatically evaluated and ranked on a leaderboard. A further bonus unit covers building agents that play Pokemon.
It assumes basic knowledge of Python and of large language models, and is free to take online. A small typo or grammar fix can be submitted directly as a pull request, while a proposed new unit is first raised and discussed as an issue.
It is written mainly in MDX and licensed under Apache-2.0.