Retrieval past the toy example: chunking that holds up, retrieval that returns the right thing, and the failure modes that only show up once real documents are involved.
For engineers building retrieval systems that have to hold up against real documents — not demo-only pipelines.
8 parts. Start at Part 1.
Part 1 — Why AI Gets Things Wrong
Beginner
Why models give fluent wrong answers: frozen knowledge and no live system access.
Part 2 — What RAG Is and Why It Works
Beginner
What retrieval-augmented generation actually is, and how it grounds answers in real data.
Part 3 — How RAG Works — The Complete Pipeline
Intermediate
Ingestion, retrieval, augmentation, and generation — and the engineering tradeoffs in each stage.
Part 4 — Chunking, Retrieval, and the Decisions That Break RAG
Intermediate
Chunking strategies, retrieval approaches, and the decisions that break RAG on real documents.
Part 5 — Build a RAG System in Practice
Intermediate
Four document shapes, four failure modes, and the decisions each one teaches.
Part 6 — RAG, Fine-Tuning, or Long Context?
Intermediate
Choosing between RAG, fine-tuning, and long context when cost, latency, and accuracy all matter.
Part 7 — Your RAG System Is Wrong. Here's How to Find Out Why.
Intermediate
Evaluation, metrics, and the diagnostic habit that finds why a RAG system is wrong.
Part 8 — RAG in Production — What Breaks After Launch
Advanced
Why production RAG drifts and quietly fails — and the discipline that prevents it.