My research and academic interests lie at the intersection of generative modeling, scientific machine learning and trustworthy AI. At CMU I work on physics-informed diffusion models for PDEs and cosmology. Earlier work at IIT Hyderabad covers explainable models and transferability estimation for ensembles. Day to day at Adobe I work on LLM evaluation and agentic systems and I’m always on the lookout for the next challenge in computer vision. Feel free to explore my work and reach out for collaborations or discussions!
AI and PDE Workshop, ICLR, 2026
Tanmay Garg, Thanh Ngoc Pham, Barnabás Póczos, Aarti Singh
A mesh-free, physics-informed conditional diffusion framework for multi-modal PDEs, covering quantum systems, wave propagation and nonlinear dynamics.
Human-Centric Representation Learning Workshop at AAAI 2024 and Deployable AI Workshop at AAAI 2024, 2024
Tanmay Garg, Deepika Vemuri, Vineeth N. Balasubramanian
The study introduces a concept learning framework, integrating an explanation generator via adversarial training, enhancing model interpretability for visual classification.
International Conference on Computer Vision 2023 (ICCV 2023), 2023
Vimal K B, Saketh Bachu, Tanmay Garg, Niveditha Lakshmi Narasimhan, Raghavan Konuru, Vineeth N. Balasubramanian
The study introduces OSBORN, a novel metric for assessing the transferability of model ensembles to downstream tasks, considering image and task disparities as well as cohesiveness between models