At Metro Infrasys, I built an end-to-end computer vision pipeline using YOLOv8s to detect and track vehicles and estimate their speed. It integrated with the company’s ANPR system to capture number plates of overspeeding vehicles and log violations in MySQL.
I also contributed to an AI-powered Automatic Vehicle Classification system using OpenCV and deep learning, categorizing vehicles in real time to support Intelligent Transportation Systems and traffic analytics.
On my Brahmaputra Erosion Risk project, I built a geospatial machine learning pipeline using 36 years of Landsat imagery to forecast riverbank erosion hotspots. I trained and calibrated a classifier, then deployed it in a full-stack app with a FastAPI backend and a React, TypeScript, and MapLibre frontend.
I also developed Coupon Ace, a mobile app using Flutter and Firebase that applies the Gemini API to extract and classify coupon data. Its recommendation system ranks offers by savings and expiry conditions, with real-time notifications.
