Machine learning for molecules, materials and proteins.
Simulation and machine learning for challenging scientific problems. Machine learning forcefields, sequence models for antibody engineering, small molecule property prediction for drug discovery. Currently an ML Engineer in the AI Sim team at SandboxAQ. Previously AbCellera, Cambridge and UCL.
Selected work
All work ↗Machine learning potentials for carbon
Method Gaussian approximation potentials fitted to van der Waals-inclusive DFT and run in LAMMPS System Carbon from graphene and diamond to the liquid, amorphous networks and free clusters Result Density, not temperature, decides whether an annealed carbon network graphitises 2025Generative language models for small molecules
Method Character-level LSTMs over SMILES, trained from scratch, conditioned on indication System ChEMBL, late-stage and approved drugs across the 50 most common indications Result Conditional generation from an indication label, with the failure modes separated out 2022–24Antibody selection and engineering workflows
Method Structure-based simulation and ML scoring over high-throughput screening data System Therapeutic antibodies, from early screening through to candidate delivery Result Deployed across more than thirty clinical targets 2021Machine learning potentials for complex aqueous systems made simple
Method Committee neural network potentials with query-by-committee active learning System Six aqueous systems, from ions in solution to nanoconfined water and an oxide interface Published Proc. Natl. Acad. Sci. U.S.A. 118, e2110077118 (2021)Writing
All notes ↗What a machine-learned potential actually learns
Descriptors, smoothness, and why transferability is mostly a data problem.
Jun 2026SMILES is a strange language to model
Canonicalisation, invalid strings, and what tokenisation costs you.
May 2026Getting a model into a screening pipeline
The gap between a good validation number and a decision someone will act on.