Data/ML/AIBackend
Hybrid Search
Also written as BM25, Keyword and Vector Search
Combining traditional keyword search (BM25 is the standard algorithm) with vector search based on meaning, then merging the results. Keywords catch exact names, codes and IDs that meaning-based search often misses.
Think of it like
Searching a library both by exact title and by asking a librarian for 'books like this one', then combining the two lists.
Junior or senior?
A sign someone has taken a RAG system past the demo stage, since pure vector search tends to fail on product codes and names.
Senior sounds like
Can describe a query that failed before they added it.
Ask them
“What kind of question did pure vector search get wrong in your system, and what fixed it?”