pgvector

A linear, one-concept-per-lesson path through pgvector, the Postgres extension that turns an ordinary relational database into a place to store and search embeddings, from your first vector column to a production-shaped FastAPI RAG service. 28 lessons · 3 tiers.

Prerequisites: Completion of LangChain & Agent Building Lessons 27-29, where notes.txt was loaded, split into chunks, embedded, and searched with InMemoryVectorStore. That lesson was honest about its own limit: fine for learning, not for production. This course picks up exactly there and builds the durable, production-shaped version, on Postgres.

Lessons use Google's Gemini free tier (gemini-embedding-001 for embeddings, gemini-3.5-flash-lite for chat), the same models as the other courses. Setup also needs a local Postgres with pgvector, run via `docker compose up -d` from the project root, a normal Postgres 17 with the vector extension already compiled in, exposed on localhost:5433 to avoid clashing with any other Postgres already running.

Built on pgvector, the Postgres extension this course teaches.