AI Engineer Academy
Become an AI engineer, one lesson at a time
A free, self-paced course from Python foundations to production AI systems — RAG, agents, evals, and deployment. Lessons unlock as you finish the one before, and new lessons ship over time as they're written. No account, no payment, ever — matches the studio's free-forever policy.
Apprentice
— · 150 XP to Practitioner
Phase 1 — Foundations
Python, tooling, HTTP, testing, databases, deployment, and the math you actually need.
Lesson 1
Toolchain: uv, virtual environments, project layout
🔒 Locked — finish the previous lesson
Lesson 2
Python core I: types, data structures, comprehensions
🔒 Locked — finish the previous lesson
Lesson 3
Python core II: functions, dataclasses, pathlib, file I/O
🔒 Locked — finish the previous lesson
Lesson 4
Python core III: exceptions, logging, generators, context managers
🔒 Locked — finish the previous lesson
Lesson 5
Standard-library CLI: argparse, json, csv (todo app capstone)
🔒 Locked — finish the previous lesson
Lesson 6
HTTP fundamentals: requests, status codes, JSON, httpx
🔒 Locked — finish the previous lesson
Lesson 7
Async Python: asyncio and concurrency
🔒 Locked — finish the previous lesson
Lesson 8
Pydantic: validating data
🔒 Locked — finish the previous lesson
Lesson 9
Project: concurrent fetcher with retries, timeouts, validation
🔒 Locked — finish the previous lesson
Lesson 10
Testing with pytest
🔒 Locked — finish the previous lesson
Lesson 11
Mocking HTTP, ruff, git workflow
🔒 Locked — finish the previous lesson
Lesson 12
SQL and Postgres
🔒 Locked — finish the previous lesson
Lesson 13
FastAPI basics
🔒 Locked — finish the previous lesson
Lesson 14
Project: CRUD API with Postgres
🔒 Locked — finish the previous lesson
Lesson 15
Docker
🔒 Locked — finish the previous lesson
Lesson 16
Deployment
🔒 Locked — finish the previous lesson
Lesson 17
Math: vectors, dot products, probability
🔒 Locked — finish the previous lesson
Lesson 18
Gradient descent and a neural net by hand
🔒 Locked — finish the previous lesson
Phase 2 — Working with LLMs
Calling model APIs directly: streaming, structured output, tool use, prompting, and cost.
Lesson 19
How LLMs work: tokens, context windows, sampling
🔒 Locked — finish the previous lesson
Lesson 20
First API calls: messages and system prompts
🔒 Locked — finish the previous lesson
Lesson 21
Streaming
🔒 Locked — finish the previous lesson
Lesson 22
Structured output and validation
🔒 Locked — finish the previous lesson
Lesson 23
Tool use (function calling)
🔒 Locked — finish the previous lesson
Lesson 24
Prompting as engineering
🔒 Locked — finish the previous lesson
Lesson 25
Cost, latency, and prompt caching
🔒 Locked — finish the previous lesson
Lesson 26
Project: streaming chatbot with tools and memory
🔒 Locked — finish the previous lesson
Phase 3 — Core AI Engineering
RAG, agents, evals, observability, guardrails, and fine-tuning — what employers actually screen for.
Lesson 27
Embeddings
🔒 Locked — finish the previous lesson
Lesson 28
Vector search with pgvector
🔒 Locked — finish the previous lesson
Lesson 29
Chunking and the RAG pipeline
🔒 Locked — finish the previous lesson
Lesson 30
Hybrid search, reranking, citations
🔒 Locked — finish the previous lesson
Lesson 31
Evals I: test sets and metrics
🔒 Locked — finish the previous lesson
Lesson 32
Evals II: LLM-as-judge and regression testing
🔒 Locked — finish the previous lesson
Lesson 33
Agents: the tool-use loop
🔒 Locked — finish the previous lesson
Lesson 34
Workflows vs agents, planning
🔒 Locked — finish the previous lesson
Lesson 35
MCP (Model Context Protocol)
🔒 Locked — finish the previous lesson
Lesson 36
Observability and tracing
🔒 Locked — finish the previous lesson
Lesson 37
Guardrails and prompt injection
🔒 Locked — finish the previous lesson
Lesson 38
Fine-tuning: when and how
🔒 Locked — finish the previous lesson
Lesson 39
Project: RAG app with an eval suite
🔒 Locked — finish the previous lesson
Lesson 40
Project: tool-using agent with tracing
🔒 Locked — finish the previous lesson
Phase 4 — Production & Depth
Reliability, durable workflows, open-weights models, multimodal, and a deployed capstone.
Lesson 41
Reliability: retries, fallbacks, queues
🔒 Locked — finish the previous lesson
Lesson 42
Durable workflows for long-running agents
🔒 Locked — finish the previous lesson
Lesson 43
Open-weights models and quantization
🔒 Locked — finish the previous lesson
Lesson 44
Multimodal: vision, speech, documents
🔒 Locked — finish the previous lesson
Lesson 45
Streaming UIs with the Vercel AI SDK
🔒 Locked — finish the previous lesson
Lesson 46
Capstone: deployed product and portfolio write-up
🔒 Locked — finish the previous lesson
Progress is stored locally in this browser only — clearing site data resets it. There is no login, no payment, and no tier beyond what's unlocked by finishing the lesson before it.