About Me
I am a Postdoctoral Researcher at the University of Oxford, working with Alison Noble, where I am exploring methods for Learning to Defer for healthcare, an area of human-AI collaboration. This research focuses on designing methods that learn to dynamically route cases between predictive models and a team of human experts based on the humans' historical performance, cost, and availability, and the AI's uncertainty.
I conducted my PhD at Imperial College London, studying machine learning methods for remote healthcare monitoring using data from passive in-home sensors and wearables. This work was conducted at the Translational Machine Intelligence Lab, supervised by Payam Barnaghi.
During my doctoral research, I built multimodal predictive models, fine-tuned language encoders, extracted structured knowledge from LLMs, and trained models on unreliable data. My PhD work developed new approaches for maching learning and evaluated methods on real-world, continuously collected, time-series data. One of my predictive models was deployed in the homes of people living with Dementia and estimated infection risk based on wearable data and in-home acitivity sensors.
Before this, I completed a BSc in Mathematics at UCL and an MSc in Applied Mathematics at Imperial, which were formative for my approach to machine learning research.
My PhD thesis is available here: Machine Learning and Remote Healthcare Monitoring Applied to Dementia Care.
My Interests
I am interested in all things machine learning and maths, but specifically:
- Machine learning for healthcare
- Human-AI collaboration
- The use of foundation models and LLMs in healthcare
- Remote or in-hospital healthcare monitoring
- Developing methods to ensure the reliable and safe use of machine learning systems
Outside of research, I am interested in photography, both film and digital, and hiking. I quickly pick up new hobbies, and am currently into making pizza and pasta dough, and collecting CDs.
Publications
Some of the publications I have worked on are listed below:
CV
Experience
-
University of Oxford, Oxford, August 2026 — Current
Post-doctoral Research Associate
Develop and evaluate Learning to Defer methods in healthcare, where methods learn to dynamically route cases between predictive models and a team of human experts based on the humans' historical performance, cost, and availability, and the AI's uncertainty.
-
Vesalic, London, 2024 — Current
Machine Learning Scientist
Working within an interdisciplinary feedback loop to drive scientific discovery.
-
Imperial College London, 2021 — 2025
PhD, Machine learning applied to healthcare
Researched the safe applications of LLMs to healthcare, representation learning of health data using self-supervised learning, predicting adverse health conditions using multimodal models, and learning from unreliable data.
-
Vector Institute, University of Toronto, 2024
3-month Study Placement, Machine Learning and Computational Healthcare
Supervised by Rahul Krishnan: We developed AutoElicit, a method for extracting expert knowledge from LLMs as structured priors for Bayesian learning of predictive tasks, with and without collaboration with clinicians. This included an evaluation of priors across a range of LLM capacities and providers.
This was funded through the Turing Global Fellows Award 2024.
-
Imperial College London, London, 2022 — 2025
Teaching Assistant for Machine Learning for Neuroscience
Prepare and present materials, instruct in labs, give tutorials on Pytorch, and mark coursework for the Machine Learning for Neuroscience course.
-
Investec, London, 2019 — 2020
Data and Analytics Graduate
Built pipelines to perform natural language processing tasks with PyTorch.
Education
-
Imperial College London, 2021 — 2025
PhD, Machine learning applied to healthcare
-
Imperial College London, 2019 — 2020
MSc, Applied Mathematics
-
University College London, 2016 — 2019
BSc, Mathematics
Projects
The following are some public projects I have worked on:
-
AutoElicit
This is a python module that was developed based on the work "AutoElicit: Using Large Language Models for Expert Prior Elicitation in Predictive Modelling". The module provides a framework for using large language models (LLMs) to elicit expert knowledge in predictive modeling tasks.
-
Machine Learning for Neuroscience Module
I was a teaching assistant at Imperial College London for the MSc Computational Neuroscience. This module introduces students to the application of machine learning techniques in neuroscience. It covers topics such as neural data analysis, predictive modeling, and the use of deep learning frameworks like PyTorch. The course website provides additional resources and materials for students.
-
AVT Python Package
Some of my graphing code, inspired by seaborn's API.
Other projects can be found on my GitHub.