Katarzyna Kobalczyk

I’m a PhD candidate in Professor Mihaela van der Schaar’s group at the University of Cambridge. My research sits broadly across machine learning and AI, with a focus on LLM-based systems that can reason under uncertainty, learn from limited information, and support decision-making in complex settings.

Rather than fitting neatly within a single subfield, much of my work lies at the intersection of several areas. I tend to approach ML and AI through the lens of decision-making under uncertainty, asking how AI systems can recognise what they do and do not know, learn from limited or language-based feedback, interact with humans and their environment, and reliably employ LLMs as part of broader reasoning and decision-making processes.

This website includes a research landscape mapping how my recent papers connect across different areas of machine learning. I see it as a snapshot of how I like to work: starting from questions around LLM-based systems and drawing on ideas from other fields such as probabilistic machine learning, uncertainty quantification, experimental design, Bayesian optimisation, or reinforcement learning. Across these directions, I am interested in finding connections that help us better understand and improve AI systems.

Before starting my PhD, I completed Part III of the Mathematical Tripos at Trinity College, Cambridge. Before that, I studied Mathematics and Statistics at the University of Warwick.

Outside academia, I have gained industry research experience through quantitative research internships at G-Research and Citadel, and an ML research internship at Meta.