My current research focuses on the intersection between retrieval information systems and LLM-based agentic systems. How we can design better agents to interact with IR systems—and how we can design better IR systems for agents to use—are some of the key questions I have been working on. In parallel, I am deeply interested in developing reliable methods to evaluate and benchmark agents that interact with IR systems. Previously, my work centered on improving deep learning data-efficiency and generalization through self-supervised learning and distillation, with a particular interest in enabling real-world deployment of such techniques. In my role at NeuralShift, I work at the intersection of NLP and the legal domain.