Joe Farrington

PhD student at the UKRI CDT in AI-enabled Healthcare Systems, University College London

Research

My PhD research is focused on using machine learning to support the management of blood inventory. I am particularly interested in using real-time data about the state of a hospital to improve institution-level forecasts and using reinforcement learning to recommend replenishment and allocation decisions.

I am supervised by Dr Kezhi (Ken) Li, Dr Wai Keong Wong, and Prof Martin Utley.

My work is generously supported by UKRI training grant EP/S021612/1 and the NIHR UCLH BRC.

Open source software

MDPax

MDPax is a Python package for solving large-scale Markov Decision Process (MDP) problems, leveraging JAX’s support for vectorization, parallelization, and just-in-time (JIT) compilation on GPUs. It is designed for researchers and practitioners who want to solve large MDPs but don’t want to become experts in GPU programming.

Background

I completed a BA in Natural Sciences at Gonville and Caius College, University of Cambridge in 2010 and an MSc in Machine Learning at University College London in 2019.

I qualified as a chartered accountant (ACA) in 2014 while working in external audit at Crowe, and subsequently spent four years as a forensic accountant at Deloitte.