About me

Hello, my name is Gaurav Bhatt. I am a PhD student in the CV-ML Lab at the University of British Columbia, where I work with Dr. Leonid Sigal. Previously, I served as a Research Assistant under Dr. Vineeth N. Balasubramanian at IIT-Hyderabad. My research experience include deep learning, representation learning, zero/few-shot learning, domain translations, and continual learning. Quiet recently, I have started working on safe and realiable use of LLMs and GenAI models.

I am also an affiliated researcher at the Vector Institute for AI in Toronto. At the Vector Institute, I worked as a Technical Facilitator for the FastLane program, RAG boot camps and BIAS in AI workshop, supervising machine learning teams from various companies such as RBC Data Science, Canadian Tire, Scotia Bank, FirstHX, RealSage, NextIntelligence, and Formic-AI.

In the summer of 2024, I began my internship at Amazon Science in Palo Alto, where my project focuses on counterfactual reward maximization for data driven ranking. I am collaborating with Kiran Thekumparampil, Tanmay Gangwani, Evan Sheehan, and Holakou Rahmanian.

In my free time, I enjoy playing badminton and table tennis, or listening to music. I used to be an avid DOTA2 player. :)

Recent Updates

[July 2024] Started internship at Amazon Science, Palo Alto, US.
[July 2024] Serving as a reviewer for ICLR'24, CVPR'24, ECCV'24, TPAMI, Pattern Recognition
[July 2024] Our paper Preventing Catastrophic Forgetting through Memory Networks in Continuous Detection. is accepted at ECCV 2024. (Code: GitHub)
[July 2024] Our paper TIBET: Identifying and Evaluating Biases in Text-to-Image Generative Models. is accepted at ECCV 2024. (Code: Github)
[Sep 2023] Our paper Mitigating the Effect of Incidental Correlations on Part-based Learning is accepted at NeurIPS 2023. (Code: GitHub, Slides : NeurIPS Slides)
[Aug 2023] Our paper Weakly-supervised Spatially Grounded Concept Learner for Few-Shot Learning is accepted at BMVC 2023 . (Poster : BMVC Poster )
[Oct 2022] Our paper Learning Style Subspaces for Controllable Unpaired Domain Translation is accepted at WACV 2023. (Code: GitHub)
[Mar 2022] Started working as a researcher at The Vector Institute, Toronto, Canada.
[Sep 2021] Started PhD at University of British Columbia under Dr Leonid Sigal.