How rakesh-100-resai/responsivle_ai_v1 works
Project type and purpose
• This is a full-stack Responsible AI evaluation platform for the Finance sector (Loan Approval use case).
• It provides a web UI for users to upload models/datasets, runs Python-based AI assessments (Fairness, Privacy, Safety), maps results to compliance frameworks (EU AI Act, RBI IT Master Direction), stores results, and renders reports.
File structure
• Top-level “app.js” – Express HTTP server entrypoint.
• “package.json” / “package-lock.json” – Node.js dependencies & scripts.
• “routes/” – Express routers (auth.js, assess.js, compliance.js, sectors.js).
• “utils/db.js” – Database connection module.
• Python scripts in root: evaluate_model.py, map_rbi_compliance.py – core AI evaluation & compliance logic.
• “datasets/” – sample datasets & metadata.
• Config/requirements files: requirement.txt (Python deps), proxy_updated_rbi_requirements.json (regulations), evaluation_report.json.
README for additional insights
• Tech stack: Node.js/Express (backend), Python 3.10+ (AI logic), static HTML/CSS (frontend).
• Features: user auth, model/dataset upload, automated evaluations, compliance mapping, report export (JSON, CSV, PDF).
Elements
• Labels: “Web Client,” “Express API,” “Python Evaluation Service,” “Database,” “File Storage,” “Config Store,” “External PDF/LLM API.”
• From Web Client → Express API (HTTP requests)
• Express API → Database (queries)
• Express API → File Storage (upload/download)
• Express API → Python Service (spawn process or HTTP call)
• Python Service → Express API (results callback)
• Express API → External PDF Generator / LLM API (optional)
• Green for backend services (Express, Python Service)
• Gray documents for config files