About Me

I am a Senior Research Scientist at Google DeepMind working on Autonomous Assistants.

My research is in understanding and designing large language models as open-ended collaborators—systems that support human reasoning, creativity, and decision-making in tasks that lack a clear ground truth, fixed objective function, or predefined solution space. This involves developing methods to evaluate not just model outputs, but the broader interaction dynamics: how models influence user trajectories, shape the space of possible actions, and steer outcomes over time. The work combines empirical evaluation, behavioural analysis, and system design to assess LLMs as epistemic agents, with particular attention to diversity, alignment, and semantic drift, in order to ensure they function as responsible and adaptive partners in complex tasks.