Choosing a vector database for ANN search at Reddit
• Written by Chris Fournier. • In 2024, Reddit teams used a variety of solutions to perform approximate nearest neighbour (ANN) vector search. • From Google’s Vertex AI Vector Sear
Ann
How hierarchical navigable small world (HNSW) algorithms can improve search
• HNSW uses layered graph structure to cut search hops in high‑dimensional vector spaces. • It balances speed, accuracy, and scalability, outperforming many ANN alternatives. • HNS
From Image Classification to Multitask Modeling: Building Etsy's Search by Image Feature
• Etsy launches image-based search on mobile, letting users find similar items via photos. • Feature uses machine‑learning embeddings to represent 100M+ listings in low‑dimensional