Research
My research focuses on data-driven decision making under uncertainty. I develop mathematical models for learning and decision systems, bringing together machine learning, stochastic control, optimization, and geometry to enable reliable decisions in uncertain environments. I am interested in how an agent should act on the data it has, when and how it should gather additional information before acting, and how it should make decisions in strategic settings in the presence of other agents.
Broader Research Interests
Beyond my primary research in data-driven decision-making under uncertainty, I am broadly interested in the theoretical foundations of machine learning. In particular, I enjoy studying how geometric, probabilistic, and optimization principles can be used to better understand and design learning algorithms.
My current interests include:
- Graph Machine Learning: understanding how graph topology influences information propagation, expressivity, and the fundamental limits of graph neural networks.
- Generative AI and Foundation Models: developing principled methods for efficient adaptation, representation learning, and fine-tuning of large-scale generative models through geometric and information-theoretic perspectives.
- Scientific Machine Learning and AI for Physics: designing machine learning models that respect the underlying laws of physics and geometry, enabling accurate, interpretable, and physically consistent modeling of dynamical systems.
Collaborators
I am fortunate to collaborate with researchers at NYU Tandon, Polytechnique Montréal, École Polytechnique, and Institut Polytechnique de Paris. I am always open to discussions and potential collaborations — please feel free to reach out.