Vector Databases in Production: Chroma, FAISS, pgvector

Descrizione del corso

A practical comparison of vector database options for production RAG systems. Covers embedding fundamentals (dimensions, distance metrics), Chroma (local and client-server mode), FAISS (flat, IVF, HNSW index types), pgvector (SQL + vector in PostgreSQL), Qdrant and Weaviate managed offerings, index selection criteria (recall vs speed vs cost), filtering strategies, multi-tenancy patterns, and migration between vector stores.

Cosa imparerai

• Curated from public resources. Compare and deploy vector databases for AI applications: Chroma, FAISS, pgvector, and managed solutions.

Programma del corso

1. Embeddings 101: dimensions, distance metrics, normalization (25 min)
2. ANN index types: flat, IVF, HNSW, PQ (30 min)
3. Chroma: setup, collections, and metadata filtering (32 min)
4. FAISS: index selection, GPU acceleration, serialization (35 min)
5. pgvector: vector search inside PostgreSQL (28 min)
6. Multi-tenancy and namespace isolation in vector DBs (28 min)
7. Hybrid filtering: metadata + vector similarity (22 min)
8. Vector DB migration: from Chroma to Qdrant/Pinecone (30 min)

Struttura del corso

3 sezioni, 8 lezioni, 9h 0m di durata complessiva. Il corso si segue interamente online, al proprio ritmo.

Domande frequenti