Clients
A sample of recent engagements — from method development to production simulation campaigns.
We developed an ML/MM coupling that embeds a machine-learned potential in a classical force-field environment: the chemically active region is treated with near first-principles accuracy while the surroundings run at force-field speed. The result is reactive-event simulation in biomolecular systems at a cost that makes enhanced-sampling campaigns practical.
For sorbent development in carbon capture, we computed adsorption isotherms of guest molecules in metal–organic frameworks, combining first-principles reference data with ML-accelerated sampling of framework–guest interactions — turning candidate structures into comparable, physically grounded uptake curves.
We simulated etching processes at the atomistic level to investigate the surface chemistry that governs material removal, and computed adsorption spectra of surface species — connecting simulation directly to the spectroscopy used to monitor and understand the process.
We license MACE machine-learned interatomic potentials for use in commercial molecular-modelling software, and support their validation, integration, and deployment — bringing first-principles accuracy to product-scale simulation workflows.
Supporting AI-driven discovery of sustainable magnetic materials, we provided first-principles and ML-accelerated modelling of magnetism — spin-resolved energetics and finite-temperature behaviour — helping rank candidate compositions before anything reaches the lab.