Why keyword search hits its limits
What BM25 does well, what it cannot do, and why vector similarity search complements the lexical index rather than replacing it.
Engineering log
Technical notes
Our write-ups on vector search, RAG pipelines, index performance and the internal architecture of Vela.
Selection
Featured articles
Choosing your embedding dimension
Memory cost per vector, how dimension affects recall, quantization and truncation: how to decide before indexing a large corpus.
All articles
Latest posts
Reducing the p99 latency of a vector index
Why the tail degrades before the median: efSearch, metadata filtering, reindexing, and how memory pressure shows up in high percentiles.
No match
No article in this topic
Pick another topic, or go back to the full list.