Soham Chitnis

I am a Masters student at Courant Institute of Mathematical Sciences, New York University (NYU) studying Computer Science. I graduated from Birla Institute of Technology & Science Pilani (BITS Pilani) in Computer Science with a Minor in Data Science.

My research focuses on Representation Learning, World Models, Multimodal Models. I also like to work on Applied AI in domains such as Healthcare, Remote Sensing essentially bringing the state-of-the-art methods to these domains.

At the CILVR Lab with Prof. Yann LeCun’s group, Wancong (Kevin) Zhang, I am training latent space World Models (JEPA & Hierarchical JEPA) with self-supervised learning methods like VICReg and optimal control planning with MPC. Beyond, this I am also exploring generative world models especially Diffusion based models.

At APPCAIR, I have worked on developing an LLM-based multi-agents for review generation of research papers. In my undergraduate thesis at TCS Research, I explored grounding large language models for Chart Understanding. I also investigated the impact of language-image pre-training of the visual encoder on the Chart Visual Question Answering (Chart VQA) task. I have worked on developing a deep learning-based model for cancer diagnosis using histopathology images and the generation of novel drug molecules for cancer cure using Deep Graph Generators. I have worked on Hyperspectral Images for various downstream tasks.

I am on job market and looking for Spring Internships in 2026 and Full-time roles starting summer 2026.

I am fortunate to have been guided by Prof. Ashwin Srinivasan, Prof. Tanmay Verlekar, Dr. Tirtharaj Dash at APPCAIR, Dr. Sidong Liu at Macquaire University, Dr. Manasi Patwardhan at TCS Research and Prof. Faisal Qureshi at Ontario Tech University.