Our Work

About This Project

MeetOps processes raw meeting transcripts using LLM to extract structured action items, with owners, priorities, deadlines, and routing them directly to Slack or Discord in seconds. Built on a pipeline combining language models with entity extraction.

Key Features

About This Project

Translates physical hand poses into keyboard and mouse inputs, enabling full game control without a controller. Combines MediaPipe hand landmark tracking with a custom-trained gesture classifier running entirely on CPU at sub-50ms latency.

Key Features

About This Project

Accepts freeform text and returns an emotional mood prediction in real time. Multi-class text classification model trained on labelled emotion datasets, served via a Flask REST endpoint with a mobile-optimised frontend.

Key Features

About This Project

Estimates which academic grade or program level a student would be in for any given target year. Born from personal frustration at guessing academic timelines, a clean, deployable utility with a simple interface.

Key Features

About This Project

A fast, clean tool for calculating fuel consumption and trip cost. Handles km or miles, petrol or EV, and INR or USD, designed for real-world use on mobile at a gas station.

Key Features

About This Project

Simulation of a Mumbai local rail station indicator. Designed in Proteus with an Arduino UNO controller, display cycles through stations with timing logic mimicking real rail intervals. Built when actual electronics were out of budget.

Key Features

About This Project

A compact LDR sensor PCB designed in KiCad for light-sensing and automation applications. Engineered for easy integration into various hardware projects with clean signal output and minimal component count.

Key Features

About This Project

A platform for renting bicycles and managing bookings seamlessly. I originated the concept, led end-to-end planning and UI design, and collaborated on backend development with a focus on clean UX.

Key Features