Skip to content
Enigma Campus

Build-along tracks

Learn AI engineering by building the real thing.

Free, step-by-step tracks that rebuild real AI systems, one commit at a time. You write the code; every lesson shows you how to check it.

Start the RAG track Browse tracks
Open nowFirst track

Build a RAG system with an evaluation set

Rebuild the ESRS-to-XBRL tagger step by step: write the evaluation set first, then retrieval, reranking and cited answers, measuring every change.

  • Intermediate
  • 9 lessons
  • 1 h 59 min
  • Python
  • FastAPI
  • PostgreSQL
  • pgvector
  • LLM APIs
See the outline
$ git log --oneline
  1. 3f9a1c2docs: describe the task and how it's scored
  2. 8b04e7deval: add 20 hand-tagged disclosures
  3. c51d0aaeval: score a keyword baseline
  4. e2a7f90ingest: chunk the ESRS taxonomy
  5. 1d6c3b4retrieval: embed chunks into pgvector
  6. 9fe2d18retrieval: rerank the top 20

How it works

Every lesson ends on a commit

You're never left wondering whether your code is right: the reference is one diff away.

  1. Step 1

    Read the why

    Each lesson starts with the problem and the decision, before any code.

  2. Step 2

    Build it yourself

    You write the code in your own repository, one small step at a time.

  3. Step 3

    Check the commit

    Every lesson ends on a real commit. Compare your code with it.

  4. Step 4

    Measure the change

    Run the evaluation set and see whether the number moved.

Tracks

What's coming next

All tracks
Coming soon

Build a document intake agent

Read invoices and forms, extract the fields, and hand the unsure ones to a person.

  • Intermediate
  • 4 lessons
  • 57 min
Coming soon

Ship a full-stack AI product

Wrap a working AI system in a product: API, interface, auth and deployment.

  • Advanced
  • 2 lessons
  • 38 min

FAQ

Questions

What does it cost?
Nothing. The tracks and their repositories are free and open.
Who are the tracks for?
Developers who can write Python and want to build AI systems that hold up in production, not just in a demo.
Do I need an account?
No. Progress is saved in your browser, so it stays on this device.
Where does the material come from?
Each track rebuilds a system Enigma Matrix built and measured. The case study shows the finished result; the track shows how to get there.

Build something that works, and prove it.

Start with the first lesson: why the evaluation set comes first.

Start lesson 1