AI consultant staffing matcher
An end-to-end retrieval pipeline for matching consultants to client engagements based on skills, experience, and project fit.
Staffing the right consultant on a project relies on tacit knowledge that is hard to surface across a large pool of CVs and past engagements. Keyword search misses semantic overlap; manual review is slow and inconsistent.
Built an end-to-end pipeline: CV ingestion and parsing, structured extraction with LLMs, semantic indexing in ChromaDB, hybrid retrieval using reciprocal rank fusion (RRF) across both dense and sparse results, and an explainability layer that surfaces why each candidate was matched.
Iterated on retrieval quality using realistic staffing queries and inspected ranking explanations to confirm matches reflected meaningful skill and experience overlap rather than surface-level keyword similarity.
A working prototype delivered as a live panel demo, showing how retrieval and explainability can be combined to support staffing decisions at scale.