Overview
VeriHire is an assistive GenAI tool for technical recruiters. Upload a job description and a candidate's resume or portfolio — it uses Retrieval-Augmented Generation (RAG) with Chroma vector search and Claude AI to cross-check alignment, surface gaps, and score fit across multiple criteria.
Stack: Python · Streamlit · Claude AI · Chroma · LangChain · RAG
What It Does
- Document Ingestion — Accepts PDF resumes, LinkedIn exports, portfolio links, and plain-text JDs
- Semantic Cross-Checking — Embeds both documents and runs similarity queries across technical skills, experience depth, and culture signals
- AI-Powered Analysis — Claude AI synthesizes vector search results into a structured fit report with pass/flag/fail per criterion
- Streamlit UI — Clean recruiter-facing interface for uploading docs, reviewing results, and exporting summaries
Technical Highlights
The RAG pipeline is intentionally simple: chunk documents, embed via a sentence transformer, store in Chroma, then query with the JD as the prompt context. Claude AI does the reasoning over retrieved chunks — it doesn't see the full documents, only the most relevant passages. This keeps token usage low and latency fast even for long resumes.
Designed as a lightweight demo that any recruiter can run locally with streamlit run app.py.
Status
Open-source demo. Actively maintained.