I am a Research Associate at the University of Bristol, developing real-time Natural Language Processing (NLP) models for child mortality surveillance in collaboration with the NHS.

In 2015, I graduated from Imperial College London with an MSci in Physics, after which I worked at Goldman Sachs for five years and developed an interest in data analytics and efficient decision modelling.

In September 2020, I joined the Interactive Artificial Intelligence CDT at the University of Bristol. Following a research master’s year (2020–2021), I completed my PhD (2021–2025) with a thesis titled ‘Bridging Neuroscience and Machine Learning: EEG Classification for Parkinson’s Disease and Beyond’. My doctoral research focused on the interpretable classification of Electroencephalography (EEG) data. While centered on early-stage Parkinson’s Disease, my work also extended to other tasks such as mental workload and emotion detection through the development of neuroscience-informed pipelines.

Research Interests

My research focuses on applying machine learning to complex time series data to obtain insights through classification and forecasting. I am particularly drawn to the challenges inherent in real-world data, such as missingness, noise, and high dimensionality.

I have extensive experience with Electroencephalography (EEG) data. In one project, we developed a classification method involving a novel transformation of the EEG data, achieving high performance on early-stage Parkinson’s classification. This work was presented at the Health Intelligence workshop at AAAI 2024 link. We extended this research for the Machine Learning and Signal Processing (MLSP) 2024 conference, where we investigated combining connectivity metrics with regional statistics for classification (see projects and publications for further details).

We have also explored state-of-the-art deep learning methods for a range of EEG tasks, including emotion detection, mental workload, and eyes-open/closed classification. Our knowledge-based pipeline outperformed established deep learning architectures adapted for EEG, with the added advantage of interpretability via feature attribution. A pre-print of this paper is available here . Below is a figure from this work showing the increased alpha band coherence observed on average for participants with their eyes closed versus open:

My work also extends to other domains, such as applying data analysis and visualization techniques to atmospheric data. I have developed projects in Python and Unity to plot and visualize global climate trends over time. The animation below is from a Unity project featuring interactive 3D visualization.

My current research involves developing NLP models for real time child mortality surveillance in collaboration with the NHS.

Please see projects and publications for further details on my work!