AI Sentiment Analysis Platform - CMU & Harley-Davidson Collaboration
- Developed a Django web app using Hugging Face, PyTorch, NumPy, and pandas for automated sentiment analysis of user comments.
- Implemented a responsive UI with Django templates and CSS, applying Harley-Davidson's signature color scheme and typography.
- Containerized the application with Docker and delivered a production-ready image for Harley-Davidson's internal deployment.
Movie Recommender System
- Deployed Django application on a VM with load balancer using Docker containerization and SQLite database serving 1M users.
- Automated CI/CD pipelines using Jenkins for deployment and GitHub Actions for PR test automation using Nose2 Python library.
- Integrated collaborative filtering model with telemetry to track drift using Prometheus and Grafana, with automated model retraining.
Goal Keeper (GoalLive)
- Implemented a Streamlit-based Python application to identify user's free time slots using Google Calendar API.
- Recommended football matches, nearby restaurants (via geolocation and Yelp API), and online match streaming options.
- Scraped match schedules using Selenium headless browser (Chrome) driver and improved performance through caching.
Covi-help Mobile Application
- Developed a cross-platform react native application leveraging the Government of India open source RESTful APIs as data source.
- Integrated AWS Code Pipeline for CI/CD, AWS EC2 hosted Inversify.js RESTful web service, and Redis for caching static data for avoiding rate-limiting and AWS S3 and CloudFront CDN for storing and caching static and dynamic assets.
