RAG

In one focused session, you’ll learn the core ideas of Retrieval-Augmented Generation (RAG) and ship a minimal but real RAG app: load your docs, chunk → embed → index → retrieve → generate an answer with sources. We use a small local dataset and open tooling so you can repeat it on your laptop after class.

Prerequisites: Basic Python (functions, virtual envs), Command line comfort(conda/pip), basic understanding of LLM's.

Learning Outcomes

• Understand the RAG pipeline end-to-end (ingest → chunk → embed → index → retrieve → re-rank → generate)
• Build a minimal RAG app that answers questions with cited sources
• Know when to tweak chunk size, overlap, top-k, and prompt structure
• Save and reload your FAISS index; swap embeddings or models cleanly