Glossary
Reranking
Re-scoring retrieved candidates for relevance before answering.
What it is
A reranker (often a cross-encoder) takes the top candidates from vector search and reorders them by true relevance to the query, improving precision before the context is handed to the LLM.
Why it matters in production
Vector search is fast but approximate; reranking is slower but precise. Retrieving 20–50 candidates cheaply and reranking down to the best 5 is the standard way to lift answer quality on high-stakes queries.
In practice
Retrieving 30 chunks with pgvector, then running a cross-encoder reranker to keep the top 5 before building the prompt.
Related terms
Production AI Notes
One practical AI engineering email each week
One concept, one architecture, one project idea, and one interview question — written for developers who want to build and ship real AI systems.