Tarzan

• Developing an autonomous vehicle portable module for non-ADAS enabled cars.
• Uses an app to input images and run custom deep learning models (YOLOv8) to make decisions for car steering angle, acceleration, and braking values.
• Takes surroundings like other cars, potholes, barricades, etc., to make its decisions.
• Worked on making simulations using MATLAB with pure-pursuit modeling.

IRIS Website

• Developed an official site for the club, handling multiple concurrent real-time payments and updating entries for events.
• Utilized Razorpay, Supabase Database, and Vercel for deployment.

Real-Time Parking Management System

• Developed a system that takes in a live video feed and allocates parking based on availability.
• Uses OpenCV and Threading with YOLOv8 for real-time allocation, speeding up the process by 40%.
• Implemented using CCTV cameras and Raspberry Pi.

Doom on Web

• Achieved a 30% performance improvement in FPS and a 10% improvement on web platforms by implementing core algorithms natively.
• Leverages Ray-Casting for its core engine algorithm and uses BFS and DFS for NPC algorithms.
• Developing multiplayer and squad-up features (WIP).

Music_Gen

• Produced music from input using LSTMs in real-time.
• Utilized open-source KERN datasets for German songs to generate unique melodies based on the input provided, giving continuations of them as a result.