Designed a wearable which allowed the user to write on any surface by tracking finger movements and digitizing them. It involved Computer Vision(OpenCV) and a Deep Learing model for finger tracking and a character recognition model with a real-time implementation. Taking notes made simple- in air, on the go.

A Classification model for tweets scraped from twitter to separate tweets which are related to natural disasters and such, from phases which have similar phrases but are being used in a different context. The model uses natural language processing to distinguish between the tweets which have similar words but different connotations.

This recommender system was designed to recommend movies, manga, games, shows, anime and all other forms of media specific to the horror genre from the media database Platform. It involves the use of a user-based collaborative filtering for recommending all of the mentioned forms of media depending on the previously watched and starred shows of a user. Designed as a slightly tweaked version of a neighborhood-based recommendation model.

Implementation of a Deep Reinforcement Learning paper to solve the algorithmic trading problem of determining the optimal trading position by using Deep QN to maximise the Sharpe ratio performance. (Ongoing)

Used existing Open-source modules and a few of my own to build a smart mirror for myself to display things like the news, the weather, my tasks for the day and calendar. Integrating it with Google, Spotify and Alexa using the APIs allowed seamless user experience with other google apps like Keep and for streaming music.

Built a prototype for a miniature arcade gaming machine using Arduino, USD sensors and a pacman-ish capture the flag multiplayer game for the system in JS.