learngenai.space

For software developers

Learn generative AI by building it.

You already write software. This is a playground for the systems around LLMs: how they generate, how you wrap them, how they fail. No API keys. No lecture videos.

Live now: How LLMs Work, about 90 minutes. After it, you can explain next-token generation and wire the loop yourself.

Start How LLMs Work

Free. No account to begin. Later modules are not shipped yet.

Try this. Same mechanic as a lesson.

The model is not looking up an answer. It splits text, weighs context, then ranks the next token.

The capital of France is …

ThecapitalofFranceis

Attention puts weight on France before the next pick.

Ranked next tokens. Pick one.

The method

Understand it. Build it. Break it. Fix it.

Each live lesson is a system you can push on. The failure is designed so you can name the cause, not watch a clip of someone else debugging.

How a lesson works

You run a system. Then we change one constraint.

No lecture. You wire the path, watch it fail for a reason you can name, then repair it. Toggle the request size.

01

Prompt

02

Context window

120 / 200 tokens

03

Generate

Ready

The request fits. Generation can start.

Live

How LLMs Work

Build a correct mental model of tokens, embeddings, attention, and next-token generation.

About 90 minutes · 20 lessons · ends in a canvas challenge

  1. 01

    Tokens

    See the pieces a model actually consumes.

  2. 02

    Embeddings

    Map meaning into space, not magic.

  3. 03

    Attention

    Watch which words matter for the next pick.

  4. 04

    Next token

    Wire scores and sampling into one loop.

Start this module

Coming next

The path after Module 1. Listed so you can see the destination, not so you wait for it to start.

  1. 02Build Your First AI ApplicationSoon
  2. 03Give Your AI KnowledgeSoon
  3. 04Give Your AI ToolsSoon
  4. 05Make It Production ReadySoon
  5. 06Evaluation as a CraftSoon
  6. 07Retrieval That WorksSoon
  7. 08Context EngineeringSoon
  8. 09Memory and Long-Running StateSoon
  9. 10Fine-Tuning JudgmentSoon
  10. 11Multimodal ProductsSoon
  11. 12Agent Systems That Survive ContactSoon
  12. 13Platform and InferenceSoon
  13. 14ObservabilitySoon
  14. 15Security and AbuseSoon
  15. 16AI Product and UXSoon
  16. 17GenAI System DesignSoon

learngenai.space

Interactive GenAI learning. One live module today. No API keys. Progress can stay on this browser until you want an account.

Start How LLMs Work