EvidenceCV

• 

  REsumeCV

EvidenceCV is an AI-powered, evidence-based resume generator built to solve one of the most frustrating problems in today's job market - qualified candidates getting filtered out by ATS systems before a human ever reads their resume. Instead of asking users to write resumes from memory, EvidenceCV automatically ingests a candidate's real work - GitHub repositories, research papers, certifications, and portfolio documents - and builds a structured knowledge base from it. Using Retrieval-Augmented Generation (RAG), the system then matches that knowledge base against a specific job description, retrieving the most relevant evidence and generating a tailored, ATS-optimized resume grounded entirely in real, verifiable experience.

Under the hood, EvidenceCV uses Sentence Transformers to embed candidate documents into a ChromaDB vector database, enabling fast semantic search powered by cosine similarity. A FastAPI backend handles document ingestion and resume generation via Cohere LLM, while a React + TypeScript frontend provides a clean user experience. The final resume is rendered in LaTeX for professional, ATS-friendly output. Built in just 3 days at the eHacks Hackathon in St. Louis 2026 - where it earned 3rd place - EvidenceCV demonstrates how RAG and modern NLP can be combined to make the hiring process fairer, more accurate, and grounded in truth.

Built With

• chromadb
• cohere
• fastapi
• huggingface
• mongodb
• python
• react
• typescript
• vite