Candidate Intelligence System

Traditional applicant tracking workflows over-index on keywords and often miss semantic relevance between resumes and job descriptions. They also make it difficult for reviewers to understand why a candidate surfaced.

The system converts resumes and job descriptions into structured intelligence. It parses documents, normalises skills, retrieves candidates through hybrid FAISS dense and BM25 sparse search, reranks matches with a cross-encoder, and produces explainable scoring signals.

FastAPI routes for auth, job, candidate, and search · Resume + JD parsers · FAISS dense + BM25 sparse hybrid retrieval · Cross-encoder reranking · SQLAlchemy 2.0 async ORM + Alembic migrations · PostgreSQL + Redis + Nginx deployment stack.