enterprise-user-management-system-ai

Clone the repository, then set up three services: run npm install in the backend and frontend directories, and pip install -r requirements.txt in ml-service. Configure .env files for MongoDB URI, JWT secret, and API URLs, then start each service on ports 3000, 5000, and 8000.

Use an authMiddleware that verifies the Bearer token with jwt.verify against JWT_SECRET, then an adminOnly middleware that checks req.user.role. Apply these middlewares to protected routes like user management endpoints.

Yes, the FastAPI ML service runs independently on port 8000 and exposes endpoints like /api/ml/classify-ticket. The React frontend or Express backend calls it via HTTP using the ML_SERVICE_URL environment variable.

Connection failures usually come from an incorrect MONGODB_URI in backend/.env or MongoDB not running locally. Use mongoose.connect with serverSelectionTimeoutMS set to 5000 and verify the database service is active before starting the backend.

The /api/ml/detect-burnout endpoint sums weighted factors: hours worked over 50 adds 0.3, overtime over 10 hours adds 0.25, missed deadlines over 2 adds 0.2, and high task completion adds 0.15. Scores above 0.5 trigger recommendations like reducing workload.