1. Paper 'LangDAug: Langevin Data Augmentation for Multi-Source Domain Generalization in Medical Image Segmentation' accepted as a poster at ICML 2025.
2. Paper 'Cycle consistent twin energy-based models for image-to-image translation' published in Elsevier's Medical Image Analysis journal.
Research Experience
1. Working at the Representation Learning Lab, IISc, under the supervision of Dr. Prathosh AP.
2. Interned at the Vision and Learning Lab, University of Alberta, with Dr. Li Cheng in the summer of 2023, working on building a multi-modal human motion dataset and exploring monocular SLAM for eye-tracking pose estimation.
Education
Completed B.Tech. in Electrical Engineering from IIT Kharagpur in 2024.
Background
Research interests include representation learning, robustness, interpretability, and multimodal intelligence. Currently a Project Associate I at the Department of Electrical Communication Engineering, IISc Bangalore, focusing on exploring the effectiveness of mechanistic interpretability methods like k-sparse autoencoders (kSAEs) for adding text-based and continuous 3D controls to text-to-image diffusion models.
Miscellany
Broadly interested in problems at the intersection of representation learning, robustness, interpretability, and multimodal intelligence. Currently looking for research assistant positions or long-term internships at machine learning and visual computing research groups.