Back to projects
AI EngineeringCohort 2
Retrieval Debugger
A trace viewer that shows why a RAG pipeline returned what it did - chunk scores, rerank deltas, and the query rewrite at each hop.
Retrieval Debugger — cover
- Side-by-side diff of pre- and post-rerank ordering
- Per-chunk score attribution
- Replays a production trace against a candidate config
Built during Cohort 2 to answer a question the team kept asking in incidents: why did retrieval pick that?
Built by