With global annual spending on prescription drugs reaching $1.3 trillion this year, the high cost of new pharmaceutical drugs is often a key political issue in Canada and around the world. Austin Tripp aims to help reduce the high research and development costs associated with new drugs by using machine learning (ML) and artificial intelligence (AI) to improve the success rate of drug testing. Building on his BASc Nanotechnology Engineering at the University of Waterloo, Austin is pursuing his PhD at a unique intersection of ML and chemistry. By computationally predicting more accurate chemical synthesis of drug candidates, fewer unsuccessful drug candidates could be trialed and millions of dollars could be saved.
His PhD research in the Department of Engineering at the University of Cambridge investigates how AI and decision theory can be used to select potential drug candidates more successfully. Unlike existing drug discovery approaches where only candidate molecules with high predicted success are tested further, Austin is developing new Bayesian models to not only predict how effective a drug could be, but also how uncertain the algorithm is of its own prediction. By exploring the computer’s uncertainty about potential drug candidates, entirely new classes of drugs potentially could be discovered that would not otherwise be chosen in the old “risk-averse” system. Utilising cutting-edge techniques like “deep generative models”, Austin is designing and testing better Bayesian models using a variety of simulated drug discovery tasks.
Austin is an outstanding young Canadian who speaks four languages. He is passionate about making a difference in the world through his commitment to engineering. He has already published a first-author paper in NeurIPS, one of the top international ML conferences. With his strong administrative skills, Austin has also organised the Cambridge Machine Learning reading group and co-organised a NeurIPS workshop. He has pursued internships at Data61, ContextLogic (Wish), NVIDIA, and Harvard University where he developed ML models for nanomaterials, information filtering techniques, neural networks for video game character animation, and algorithms for predicting liquid mixture compositions.
After completing his PhD, Austin plans to continue doing research in the intersection of AI and the natural sciences. With the technological landscape rapidly changing, it is unclear where this will take him, but academia is a good bet, hopefully back home in Canada!
Austin was awarded the Copper Street Capital LLP Scholarship 2022 – 2023
