How miguelmf-coder/matchkey-genai-maverick works
Project type and purpose
• MatchKey is a full-stack AI-driven job-matching platform.
• The README and file tree show a Streamlit frontend, a FastAPI backend, a PostgreSQL database, and multiple AI “agents” (OCR, skills extractor, HR Copilot, matching/co-teaching engines, scrapers).
File structure
• Top level:
• docker-compose.yml orchestrates three services: frontend, backend, db
• backend/ holds Dockerfile, requirements.txt, and the FastAPI app
• frontend/ holds Dockerfile, requirements.txt, app.py (Streamlit), utils, and role-based pages
• docs/ holds planning files
• backend/app:
• routers/: FastAPI endpoints (auth, candidates, companies, jobs, matching, copilot)
• services/: AI agents and utilities (OCR, scraping, matching, hr_copilot, notifications, recommendations)
• models/, db/: SQLAlchemy models, session, init/seed scripts
• schemas.py for Pydantic schemas
• frontend:
• app.py + utils.py as entry point
• candidate/ & company/ modules with Streamlit pages and call_ai integrations
• Config files:
• backend/Dockerfile, frontend/Dockerfile, docker-compose.yml
README for additional insights
• Architecture summary:
• Frontend: Streamlit (port 8501)
• Backend: FastAPI (port 8000)
• DB: PostgreSQL
• MCP Multi-Agent system comprising Document Parser, Skills Extractor, HR Copilot, Matching Engine, Co-Teaching, Scrapers
• Lists main REST routes and DB tables