Better retrieval: hybrid search, reranking and citations
Comes up in: AI engineering, AI system design
This concept is being written. It gets the full five steps — explanation, predict, interview angle, practice with tests, and review cards. Here is exactly what it will cover, so you can study it elsewhere in the meantime.
What it covers
- keyword (BM25) + vector hybrid
- reranking models
- query rewriting
- contextual retrieval idea (adding chunk context — cite Anthropic post, 2024)
- citations and grounding
- when long context beats RAG
Words you will know
What you will practise
- improve recall@5 on a provided eval set
Interview questions this answers
- What is hybrid search and why does reranking help?
- When would you skip RAG and use a long context window instead?