About Kavit Tolia
I spent 12 years in asset management, working with quantitative models, data, uncertainty and investment portfolios.
Over time, I found myself increasingly drawn to the mathematics and mechanics underneath the decisions: how models represent the world, what information they capture, and where their assumptions begin to break down.
That curiosity eventually led me to step away from finance and pursue an MSc in Artificial Intelligence and Machine Learning in Science. I am now a PhD researcher at Queen Mary University of London, funded by EPSRC, working on autonomous optical networks: how reinforcement learning and generative AI agents can translate an operator’s plain-language intent into the decisions that run a fibre network.
I enjoy trying to understand technical ideas from first principles and finding ways to make their underlying structure easier to see. I do not think difficult subjects need to be made artificially simple, but I do think they can often be explained more clearly.
My current interests include reinforcement learning, scientific machine learning and inverse problems, and increasingly the physics of optical fibre, which is new to me. More broadly, I am interested in what models can learn from data, what remains unidentifiable, and how visualisation can help expose both the strengths and limitations of a model.
I am still learning, and I expect my understanding of many of these subjects to change. Part of the reason for building this site is to document that process.