Learning path

Start at the top if you are new. Strong Python developers can skip ahead to math or jump to Agents after LLMs — but do not skip evals.

1. Stage 01

   Getting Started

   What autonomous agents are, how this academy works, and how to think in loops.

   Open first lesson

2. Stage 02

   Python

   Simple Python from zero: names, lists, functions, files, JSON, errors, tests, and the loop every agent uses.

   Open first lesson

3. Stage 03

   Mathematics

   Simple math from zero: lists of numbers, scores, chance, entropy, and the formulas agents actually use.

   Open first lesson

4. Stage 04

   Machine Learning

   Simple ML from zero: examples, splits, loss, ranking, and the loop you use to judge an agent.

   Open first lesson

5. Stage 05

   Neural Nets & Transformers

   Simple transformers from zero: tokens, attention, decoding, and the stack every agent call runs.

   Open first lesson

6. Stage 06

   Large Language Models

   Simple LLMs from zero: chat APIs, tokens and cost, JSON actions, and when not to call a model.

   Open first lesson

7. Stage 07

   Prompting

   Simple prompting from zero: four parts, examples, tags, injection, evals, and the OS an agent actually reads.

   Open first lesson

8. Stage 08

   Tools & Function Calling

   Simple tools from zero: schema, dispatch, idempotency, sandboxes, MCP, and permissions your runtime actually enforces.

   Open first lesson

9. Stage 09

   RAG & Memory

   Simple RAG from zero: chunks, cosine, hybrid search, citations, agentic retrieve, and memory that is not one vector soup.

   Open first lesson

10. Stage 10

    Agent Architectures

    Simple agent loops from zero: six parts, ReAct, plan-and-execute, reflection, typed state, HITL, jobs, and when not to agent.

    Open first lesson

11. Stage 11

    Multi-Agent Systems

    Simple multi-agent from zero: split only when tools and checks diverge, typed handoffs, supervisors, debate, swarms, and how teams fail.

    Open first lesson

12. Stage 12

    Evals & Safety

    Simple evals from zero: golden properties, tool tests, judges you measure, traces you replay, injection allow-lists, and a spec that can say no.

    Open first lesson

13. Stage 13

    Production Agents

    Simple production from zero: gateway and jobs, traces you can operate, caps and queues, secrets, CI gates, and kill switches you have practiced.

    Open first lesson

14. Then build

    Weather agent → ReAct research → RAG support → multi-agent team → ops agent.

    Open projects