Research Interests

Video Representation Learning: My research uses representation learning approaches to extract robust, high-level representations from video data. Recently, I specialized in leveraging diffusion transformers to learn transferrable representations suitable for transfer learning and downstream video understanding tasks.

Video Editing & Motion Transfer:  I work on developing training-free techniques for video manipulation and motion transfer between subjects. This includes designing deep learning models' latent space for video editing, motion retargeting, drastic shape editing and temporal consistency preservation in edited videos.

Fairness:  My research also explores fairness and bias mitigation in computer vision systems, particularly in image analysis applications. I focus on developing algorithms and frameworks that ensure equitable performance across different demographic groups while maintaining high accuracy.