Micro-Course · AI Engineering
RAG Fundamentals
Intermediate 33 min 5 Bytes AI Engineering
Build a real retrieval-augmented generation pipeline — chunks, embeddings, retrieval, reranking, evals.
What you'll cover
- RAG in 10 minutes — The smallest possible retrieval-augmented generation pipeline you can ship.
- Vector DB basics — What an embedding store actually does, and when you don't need one.
- Semantic chunking — Why naive text splitting wrecks RAG quality, and what fixes it.
- Rerankers, explained — Cross-encoders, cohere rerank, recall@k vs precision@k.
- Evals for LLM apps — Pairwise, rubric, golden sets.