Developed a comprehensive Personal Finance Dashboard, integrating advanced data visualization techniques, which allowed for tracking expenses and savings, contributing to an increase in monthly savings.
Developed 'OTT', a travel application using React Native, enabling users to book cabs and travel packages, with separate interfaces for users and cab drivers, and integrated Google APIs for live driver tracking.
The AI Image Retrieval project is an image search application designed to provide efficient and accurate retrieval of images based on user queries. The user query can be of Text, Image, Sketch or a combination of Text & Image.
Developed a twist on classic Tic-Tac-Toe where every three moves, the oldest one vanishes. Discover the gameplay, strategy, and logic behind Tic-Tac-Toe Vanish.
Built a complete pipeline for performing VQA on Image Query Answering in Amazon ML Challenge 2024. Employed Multimodal ML models and language processing to generate accurate query responses. Achieved an All-India rank of 106 and an F1 score of 0.55
Implemented a machine learning solution for automated answer sheet grading using advanced image processing; boosted grading efficiency and enhanced evaluation precision, handling over 600 student sheets per hour.
Employed Physics-Informed Neural Networks to solve Partial Differential Equations, beginning with a standard PDE to validate the methodology. After establishing the effectiveness of PINNs on simpler equations, we extend the approach to solve the Time-Dependent Schrödinger Equation, a core equation in quantum mechanics.
Developed SynthData, an npm package enabling versatile mock data generation for testing and prototyping. Features include random data generation, structured datasets, seamless integration, and customizable numeric data.
Built and fine-tuned a machine learning model for music genre classification using the GTZAN dataset, achieving a 90% accuracy rate in identifying 10 different genres from audio tracks, significantly enhancing user experience and engagement. Enhanced music genre identification with over 90% accuracy using advanced techniques, aiding in the improvement of music recommendation systems.