Every year, G-Research runs a number of different PhD prizes in Maths and Data Science at universities in the UK, Europe and beyond.
Each prize is worth up to £10,000 and is open to final or penultimate year PhD students at specific universities, working across areas including Machine Learning, Quantitative Finance and Mathematics.
We’re pleased to announce the winners of this PhD prize, run in conjunction with Columbia University.
“My research centres on a fundamental question in robotics: how can robots better understand the world and interact with it?
“As a final-year PhD candidate in Computer Science at Columbia University advised by Prof. Yunzhu Li, my PhD research advances the frontier of representation learning for robotics by introducing a wide spectrum of structured representations that enable critical downstream applications, including stable real-time action-conditioned video generation, model-predictive control, generalisable imitation learning and task planning.
“My research has been broadly recognised by the robotics community, including an oral presentation at CoRL 2024, finalist recognition for the CoRL 2023 Best Paper and Best Student Paper Awards, an ICRA 2024 Best Paper Award, multiple workshop best paper awards and selection as a 2025 Qualcomm Fellowship Finalist.”
“My research focuses on Trustworthy AI, specifically the development of reliable agentic systems and data-driven algorithms.
“While my work has already been adopted by industry leaders like Microsoft and Barracuda Networks and featured in Forbes and MIT Technology Review, my current mission is to bring enterprise-grade AI to the masses.
“I truly believe the future of work lies in agentic systems that are not just powerful, but provably dependable.”
“I work on trustworthy multimodal AI, with a focus on making systems more interpretable, reliable and safe in complex real-world settings.
“My research develops methods that expose and shape model rationales, use code and tool-calling to make multimodal reasoning more explicit and auditable, and evaluate decision processes rather than final answers alone.
“Across this work, I aim to show that structure and transparency need not come at the expense of capability and that systems which reason through human-readable evidence, executable programs, and verifiable intermediate steps can be both more powerful and more trustworthy. “
