// hi, I'm
Class 12 student from Bangalore. I train models, ship products, and refuse to wait until I have a degree to do real things.
AI-powered urban pollution monitoring platform built for the Coding for Climate competition. Real-time AQI dashboard pulling live city data, a carbon footprint calculator with emission factors, an AI generated air quality overview, and a YOLOv8 multi-class computer vision module that detects smoke, garbage, vehicles and fire from images. Trained on ~52,000 samples across dual T4 GPUs on Kaggle notebooks.
It's difficult to find entrar assignments with ease, especially if you're in a hurry. They are hidden behind multiple pages and dropdowns. Pretty Entrar Dashboard comes in and saves you from all the trouble. Simply login with your credentials and find all assignments instantly! No hunting needed anymore.
A simple webapp that classifies the likeliness of a pothole forming on a road via a logistic regression to perform multi-variable classification. Trained on data scraped off of government portals and datasets. A cron job also gets the /health endpoint to prevent cold-starts
Assignment Tracker is a lightweight internal web app built with FastAPI and Google Sheets as a backend database. It lets a team create, assign, and track tasks with AI-assisted priority suggestions powered by the Gemini API. Features include a mobile-friendly interface, colour-coded priority badges, inline status updates, and a view for each team member's assignments. Built entirely from scratch: backend, frontend, and Google Sheets integration.
Signal processing pipeline that uses IMU sensor data to detect squat reps, returns a pandas dataset rep-wise and and classifies fatigue patterns across a workout set. Built with pandas, and numpy.
I'm a Class 12 student in Bangalore obsessed with building AI systems that solve real problems, not just Jupyter notebooks that never leave my laptop.
Right now I'm focused on computer vision, deployed ML pipelines, and full-stack AI products. I also enjoy graphic design and photography.
I believe the best way to learn is to ship things, break them, and ship again. Everything in my portfolio is live or on GitHub. No vaporware.
Find me on GitHub.
03. what's next
open to collabs, internships, and interesting problems. reach out.