Smart Screen

A resume screening tool built for HR teams drowning in applications. RAG Fusion’s multi-query retrieval re-ranks candidates across 100+ resumes against a job description, maintaining 90%+ precision while delivering top-5 candidate matches in under 3 seconds.

Before: a recruiter spent 30 minutes manually scanning a batch of CVs. After: under 1 minute, with an explainable ranking that cites the specific resume passages that drove each match.

Built with Streamlit for the interface, LangChain for the retrieval pipeline, and PyTorch for the embedding model. The multi-query approach (generating several variants of the job description query) substantially outperforms single-query RAG on precision at k=5.