He is currently working on the development of STRIAD solutions as part of the technology development team, particularly focusing on the INNOVATE UK funded CESIUM project. This project aims to develop better methodologies for identifying risk of child exploitation using data science and machine learning. Ezra is particularly interested in the development of model-agnostic algorithmic transparency and explainability techniques for elucidating the decision-making processes of machine learning models. This is to provide practitioners with explanations and insights they can use to effectively audit, verify, contextualise, and potentially falsify, a model’s outputs, thus enabling human-focused machine-assisted decision-making, as opposed to prescriptive autonomous decision-making.
Ezra completed his MSci in Physics at the University of Cambridge, where he also earned a BA in Natural Sciences. His master’s thesis was based on using deep learning and graph neural networks to develop models for complex physical dynamical systems in an unsupervised way.
He has also conducted research projects, involving deep learning, computer vision and large-scale Monte Carlo simulations, at the University of Tokyo, the California Institute of Technology, the Rutherford-Appleton Laboratory, and the Cavendish Laboratory.