Step 01 / 04

You are going to build AI systems

Module map

Most people start learning Generative AI by memorizing vocabulary: LLM, RAG, embeddings, agents, attention, evaluation. That feels productive — until you try to design a real feature and realize the words never turned into judgment.

We take a different path. You will build small systems first, poke at them, break them on purpose, and only then name the pieces. By the end of Module 1 you should be able to explain what happens between a user’s question and a model’s answer — in plain language.

The loop for this whole course: Understand it → Build it → Break it → Fix it. Every major idea gets that treatment.

Who this is for

You do not need a machine-learning background. If you are comfortable with functions, APIs, databases, and HTTP, you already have the right foundation. Think of an LLM like a specialized remote service — powerful, imperfect, and only as useful as the application you wrap around it.

  • We prioritize mental models over formulas.
  • Simulations are educational, not production benchmarks.
  • You will leave able to say why a design choice exists — not just what it is called.