popular Vector Indexing - HNSW, IVF

vector DBs exist for one fundamental reason: to do approximate nearest neighbor (ANN) search fast at scale. Everything else — filtering, hybrid search, persistence, distributed sharding — is built around that core operation.

If you remember nothing else: vector DB = ANN index + a query layer + persistence + ops.

• 1. curse of dimensionality
• 2. The core trade-off triangle Every vector index sits on a triangle:

3. Indexing algorithms the heart of the topic

• 4. Distance metrics - You covered embeddings already, so brief
• 5. Filtering — and why pre vs post matters
• 6. Hybrid search and reranking

7. Production concerns — the FDE lens

• 8. The landscape — picking one
• 9. The customer-facing framing