Articles for people who
ship the stack
Rewrites on React, Node.js, TypeScript and AI. A translated body is shown when it exists. If it does not, you see the English text and a note that says so.
Tagged: HNSW
Five Ways to Index Vectors for RAG, Tested on a Million Wikipedia Embeddings
Five ways to index vectors for RAG tested on a million Wikipedia embeddings: flat, IVFFlat, HNSW, BM25 hybrid and DiskANN, with latency and recall numbers.
9 min readVector Search in Azure AI Search: HNSW, Profiles and a Python RAG Retriever
Build a vector index in Azure AI Search with Python: HNSW vs exhaustive KNN, profiles, embeddings, vector queries and a RAG prompt, plus tips for hybrid search.
9 min readInside Vector Databases: IVF, Trees, HNSW and Product Quantization
Inside vector databases: exact vs approximate search, IVF clusters, tree forests, HNSW layers and their parameters, and what product quantization costs you.
7 min readFinding Why a Weaviate Query Is Slow With Per-Stage Profiling
Weaviate query profiling returns a stage-by-stage timing breakdown with each response, so slow filters, disk reads, graph search or rescoring can be told apart quickly.
5 min readWe Dropped Our Vector Database: pgvector Served 40M Embeddings Fine
A migration story: replacing a paid vector store with pgvector on 40 million rows, including the backfill method, the recall mistake that hid in dashboards, and when not to switch.
5 min readBeyond Naive RAG: HNSW Memory Costs, Hybrid Search and Long Context
Why in-memory vector indexes hit a wall at scale, where exact-match retrieval fails, and how BM25 hybrids, RRF and PostgreSQL fit in a long-context world.
5 min readHow a Vector Database Works: Embeddings, Vector Search and Human Memory
A memory-based mental model for vector search: how embeddings, similarity, nearest neighbor indexes and metadata filters work inside a vector database for RAG.
9 min readTuning HNSW Indexes and Scaling Vector Search for Production RAG
Learn how HNSW's M and ef parameters trade recall, latency and memory, how to set them in Chroma, and when to scale a vector database vertically or by sharding.
4 min readHow Vector Databases Really Work: From Embeddings to Hybrid Search
Explains how embeddings encode meaning, how similarity search and indexing scale, and when hybrid search and vector databases actually fit enterprise AI systems.
8 min read
About these articles
Request a 24h estimate
Need the same stack in a production operator layer? Send the brief — estimate within 24 hours.