About Me

I am a post-doctoral research assistant at Imperial College London interested in using Bayesian machine learning and causality to drive new discoveries in biotechnology. My research so far has focused on using Gaussian processes for the experimental design of biotechnology experiments and for causal discovery. I previously did a PhD at Imperial College London as part of the UKRI Centre for Doctoral Training in AI for Healthcare supervised by Ruth Misener, Mark van der Wilk and Molly Stevens.

Projects

• Currently, I am working on a project with researchers at Imperial College London and Oxford University using Bayesian model selection for multivariate causal discovery. Check back soon for updates.
• I’ve also collaborated on a project extending Gaussian process latent variable models to the case where each data point is a weighted sum of signals, specifically for applications in spectroscopy.
• In collaboration with experimentalists from Oxford University and KTH Royal Institute of Technology in Sweden, I am working on an integrated machine learning design of experiments approach the development of lateral flow tests.
• My main PhD project focused on transfer learning Bayesian optimisation to optimise DNA molecules for use in medical diagnostics. You can find the Biotechnology and Bioengineering article here.
• I also developed a design of experiments strategy for validating or invalidating designed biomolecular networks in a small number of experiments. You can find our NeurIPS 2020 ML4Molecules Workshop paper here.
• I co-supervised a master’s student project investigating the use of sparse Gaussian processes to improve air pollution predictions in Kampala, in a collaboration with AirQo that started at the 2022 AI for Social Good Dagstuhl workshop. You can read our NeurIPS 2023 Climate Change AI workshop paper here.

Research Interests

• Bayesian optimisation and active learning – techniques and methods for learning about a black box function in an intelligent way.
• Causal discovery and active learning – how we can improve posteriors over graphs with the aim of using them in decision processes.
• Transfer learning – how we can use information learnt from one task to inform another, with the aim of reducing experimental burden.
• Biotechnology and diagnostics – collaborating with experimentalists to develop better medical diagnostics for point of care settings.
• Design of experiments – how we can use Bayesian machine learning techniques to improve experimental design for lab experiments.
• Machine learning for air pollution monitoring - how we can improve predictions of air pollution levels for improved decision making.