2024
Hierarchical Bayesian Program Synthesis
A wake-sleep framework for learning reusable neural primitives for algorithmic reasoning — MSc thesis with distinction.
↗Profile
I work on large-scale distributed training of LLMs at Graphcore. I’m also interested in programmatic representations. Previously, I did research on neural program synthesis with Prof. Mirco Musolesi in UCL’s Machine Intelligence Lab.
I did an MSc in Machine Learning at UCL and a BSc in Physics at Imperial College London. At Imperial, I worked with Dave Clements and later spent a summer researching causal discovery with Mark van der Wilk. I grew up in Budapest, Hungary.
Work & writing
Research, writing, and selected projects.
2024
Masked Image Modelling
Self-supervised Vision Transformer pre-training for downstream segmentation.
↗2024
ML@UCL
Projects and assignments undertaken during my master’s: COMP0086 Probabilistic and Unsupervised Learning (Gatsby Unit) Bayesian Model Selection Expectation-Maximi...
↗2023
Bivariate Causal Discovery with GPLVMs
Imperial UROP research on causal discovery using cross-validation and Bayesian model selection.
↗2023
Evolution of Artificial Life
An agent-based study of optimisation in gene space, coevolution, and natural selection — BSc thesis with distinction.
↗2022
Machine Learning for LHCb
As part of Physics Coursework at Imperial College London, I led the machine learning team analysing decay product distributions from the Large Hadron Collider beauty (LHCb) expe...
↗2022
Quantum Computation: Hopes and Struggles
Imperial College article exploring the promise and challenges of quantum computing — Quantum Article Prize winner.
↗2022
Project Svarog: Solar Sail Mission at ICSS
Part of the orbital mechanics team at the Imperial College Space Society, I contributed to Project Svarog — a CubeSat with a solar sail set on a trajectory to interstellar space...
↗Background
Experience
-
Present
Graduate Machine Learning Engineer
Graphcore · Cambridge
-
2025
ML Research Engineer Contract
Ascentra Labs · London
LLM integration, structured generation, and comparative model evaluation for a survey analysis product.
-
2024
Postgraduate Researcher
Machine Intelligence Lab · UCL
Neural program synthesis and algorithmic reasoning with Prof. Mirco Musolesi.
-
2023
Research Intern
Imperial College London · Mark van der Wilk
Causal discovery and Gaussian Process Latent Variable Models with Mark van der Wilk.
Education
-
2023—24
MSc Machine Learning Distinction
University College London
Thesis: Hierarchical Bayesian Program Synthesis for Neural Algorithmic Reasoning.
-
2020—23
BSc Physics with Theoretical Physics Honours
Imperial College London
Thesis: Evolution of Artificial Life: Investigating Optimisation in Gene Space.