TrackML

TrackML is a full-stack tool for tracking, managing, and analyzing machine learning models. It supports organizing models, extracting insights using AI, and integrating with external APIs for automation and enrichment.

Key Features

• Track details of ML models such as name, type, developer, and parameters
• Organize models using tags and status labels
• Search and filter models by different attributes
• Visual dashboard showing model stats and insights
• Dark mode for better UI experience
• Auto-fill model info from HuggingFace API
• Compare models and generate insights using Google Gemini AI
• Mobile-responsive design
• Secure login and authentication
• Cross-origin request support
• Ready-to-deploy on AWS EC2

Architecture Overview

Frontend

• Built using React 18+ and Vite
• TypeScript used for safer code
• TailwindCSS for styling
• Routing with React Router v6
• Global state handled via Context API
• Axios used for API communication

Backend

• Flask powers the REST API
• SQLAlchemy ORM for database operations (SQLite/PostgreSQL supported)
• JWT used for authentication
• Integrates Google Gemini and HuggingFace APIs
• Flask-CORS enables secure cross-origin communication
• Written in Python 3.8+

Infrastructure

• Hosted on AWS EC2
• Optional use of AWS RDS for databases and S3 for static assets
• Domain and SSL managed through AWS Route 53 and ACM

Nginx Configuration Summary

Reverse Proxy

• Routes requests to backend and frontend
• Supports HTTPS termination
• Handles headers like X-Real-IP and X-Forwarded-For
• Configurable for WebSocket and subdomain routing

Caching

• Uses disk and memory caching
• Supports API and static file caching
• Allows custom cache control rules
• Can bypass or purge cache as needed

Performance

• Enables compression via GZIP or Brotli
• Optimized for large headers and payloads
• Manages keep-alive connections and worker events

Security

• Implements rate limiting using IP zones
• Adds HTTP security headers to prevent XSS, CSRF, and clickjacking
• Configures SSL with strong ciphers and modern protocols

Monitoring

• Logs requests, errors, and performance metrics
• Tracks response times, error rates, and cache efficiency

Deployment

• Local setup requires Python 3.8+, Node.js 16+, and API credentials
• EC2 deployment involves setting up services, Gunicorn, and Nginx
• Frontend is built and served using Nginx with backend proxied on port 5000

API Summary

• Auth endpoints for registration, login, and logout
• Model endpoints for create, read, update, delete, search, and autofill

Project Structure

• Backend: organized into models, routes, services, and config
• Frontend: includes components, pages, services, and types
• Deployment: contains Nginx and systemd setup files

Troubleshooting

• Ensure CORS is configured properly in the backend
• Verify environment variables including API keys and database URLs
• Make sure external services (e.g., Gemini, HuggingFace) are accessible

Licensing and Credits

• Licensed under MIT
• Uses APIs from Google Gemini and HuggingFace
• Hosted using AWS infrastructure