About
I’m Leon Ross. I work at the intersection of machine learning, biology, and markets.
I’m currently doing an MPhil in Computational Biology at the University of Cambridge (Fitzwilliam College), focused on machine learning for perturbation biology: models of how cells respond to genetic and chemical perturbations, and how to use them to decide which experiments — and which drug targets — are worth pursuing. Right now that means building an active-learning harness for perturbation screens and preparing a PhD application to the Wellcome Sanger Institute.
Before Cambridge I read Mathematics at the University of Bristol, graduating with First Class Honours, and then spent time as an investment banking analyst at J.P. Morgan Cazenove. Banking taught me financial modelling and company analysis, and left me with a lasting interest in the business side of science: how early-stage research becomes a product, and who captures the value when discovery gets cheaper.
Those two threads — methods and markets — are what I write about. My working view is that most perturbation experiments will eventually be predicted rather than run, and the interesting questions are which predictions to trust, which experiments still earn their cost, and what that does to the economics of drug discovery. Notes on both the papers and the companies are on the writing page.
Outside of work I’m usually reading, running, or arguing about whether a given techbio company has a moat.