RAG & Agents
Retrieval is how you give a model knowledge that wasn’t in training. Agents are how you give it actions. This module is end-to-end practical: every lesson ends with a working agent or RAG pipeline you can run.
0 / 4 lessons ~59 min total
Module capstone — build it
RAG over your own GitHub repo — deployed by Sunday night Index your own code. Ask 'where is X handled?' Get cited answers. Live at a URL your friends can hit.
Intermediate · One weekend · Runs on your laptop
Pick a repo (yours or a favorite open-source project). Chunk and embed every file with code-aware splitting. Build hybrid retrieval (BGE + BM25 + RRF) plus cross-encoder reranking. Wire to Claude. Add a Next.js chat UI. Deploy to Vercel + Qdrant Cloud (both free tier). Share the URL.
Tools you'll use BAAI/bge-small-en-v1.5 (embedder) BAAI/bge-reranker-v2-m3 Qdrant Cloud (free tier) Claude 3.5 Sonnet Next.js + Vercel