Updated for 2026 · v0.1 Aurora
Vector databases
for AI engineers.
The missing manual for similarity search. Learn how embeddings, ANN indexes and modern vector engines come together to power retrieval, RAG and recommendation systems.
Engines covered
Six production-grade vector stores
What you'll learn
From a 1536-dim vector to a production RAG system.
- How embedding models turn text, images and audio into vectors
- Cosine, dot product and Euclidean — when to use which
- HNSW, IVF-PQ and DiskANN: how ANN indexes really work
- Hybrid search combining BM25 + dense vectors with re-ranking
- End-to-end RAG: chunking, retrieval, generation, evaluation