- Engineered in-house large music model with 3.3B parameters, trained on diverse 5.3K+ hrs of music data- Utilized deep learning to develop a robust architecture with 50+ instrument specialized Nano models.- Integrated all deployments into the MLOps pipeline to ensure operational efficiency and compatibility.- Implemented end-to-end data collection and training pipelines for SFX model to generate foley sound.- Researched on a novel AI arranger model to compose harmonical music from solo instruments tracks.

- Researched on recommendation algorithms, focusing on user-item based collaborative filtering and content-based methods.- Implemented a hybrid of collaborative filtering (CF) and content-based (CB) approaches using stacking and blending ensemble techniques in PyTorch.- Collaborated with cross-functional teams to integrate recommendation engines into the educational platform of SATHEE. Conducted A/B testing to evaluate the effectiveness of different recommendation strategies.- Also worked as a TA for the SATHEE project, an initiative by NTA, MHRD, and IITK to assist students in preparing for IIT-JEE without coaching. Created 600+ relevant problems, essential formulae, and tips & tricks sheets for JEE aspirants to practice.