Gaurav Bhatt received his PhD in Computer Science from the University of British Columbia, where his research focused on developing robust and adaptive machine learning systems. During his doctoral studies, he worked on continual learning, representation learning, and foundation model adaptation, with contributions spanning computer vision, large language models, and multimodal AI. His recent work investigated how foundation models can be fine-tuned for new tasks while preserving critical alignment properties such as safety, truthfulness, and reliability.
Gaurav’s broader research interests lie at the intersection of foundation models, machine learning, and healthcare. He develops methods for alignment-aware fine-tuning, reward learning, continual adaptation, and preference optimization, with the goal of enabling the safe and effective deployment of AI systems in high-stakes settings, where model errors can have significant real-world consequences. His research is particularly motivated by applications in healthcare, where AI systems must remain reliable, trustworthy, and aligned with human values when supporting clinical decision-making, medical question answering, and biomedical discovery.
By integrating advances in machine learning, optimization, and human feedback, Gaurav aims to develop trustworthy AI systems that can be adapted to specialized domains while maintaining robust performance and alignment. His long-term goal is to advance reliable foundation models for healthcare that can safely assist clinicians, accelerate biomedical research, and improve decision-making in real-world clinical environments.