Hi, I'm Pranav.

I’m a soon-to-graduate PhD student and a Neuro-AI researcher at the University of Oxford, looking for postdoctoral research opportunities. My background is in engineering. I study minds as computational systems—entities that adapt and act purposefully in uncertain, non-stationary, open-ended environments. 

I’m recently interested in how minds solve hard problems under constraints, such as continual lifelong learning, hierarchical planning and reasoning, long-range credit assignment, and survival-critical computations. Brains solve these challenges under biological constraints such as local learning rules, compute-bound, resource and sample efficiency, and mortality, which inspire alternative approaches for AI. 

By drawing on neuroscience, I aim to (1) advance fundamental AI research and (2) develop tools to mechanistically interpret and characterise behaviours and learning in complex AI systems. On the flip side, I draw inspiration from AI research to (3) better understand cognition and to (4) view mental health disorders and chronic pain as disruptions in continual learning, opening paths for new treatments. 

I did my PhD on safe learning in humans and machines, where I studied safe exploration using multi-objective RL and self-preservation/homeostasis using POMDPs. I’ve had the privilege of working with Peter Dayan, Rafal Bogacz, Boris Gutkin, Flavia Mancini, Sang Wan Lee, and my advisors Ben Seymour and Ioannis Havoutis. Previously, I briefly worked on scalable distributed deep learning at Nvidia.