Scientific machine learning · medical imaging · research software
I am a Cambridge-trained physicist completing an MPhil in Data Intensive Science. In 2026, I will begin an EPSRC-funded DPhil in Healthcare Data Science at the University of Oxford's Big Data Institute.
My work sits between physical science, machine learning and software engineering, with a particular interest in reliable computational methods that remain interpretable and reproducible.
- Patch-Based Diffusion Models for CT Reconstruction — reproduced PaDIS within the LION tomography framework, adding reconstruction algorithms, experiment pipelines, tests and documentation for cloud and HPC environments.
- Interpretable FTIR Spectral Classification — my Cambridge MSci project and a NeurIPS ML4PS 2025 workshop paper, combining a physics-informed CNN ensemble with interpretable spectral contexts.
- FACTTRACE — a transparent multi-agent truth jury developed for the North Star Hackathon, where our team placed second.
- Active Alarm — an end-to-end Apple Watch movement-classification system spanning synchronised data collection, PyTorch training and on-device inference.
I primarily work with Python, PyTorch, JAX, Swift, JavaScript/TypeScript, scientific computing, GPU/TPU workflows and reproducible research tooling.