PhD in Engineering

Designing materials and structures with unprecedented performance requires constitutive models that can generalise across chemistries, microstructures and loading histories. Classical approaches rely on hand-crafted modelling assumptions (e.g. specific elastic - plastic - viscoelastic laws with prescribed internal variables) that must be re-tuned or completely re-built whenever the microstructure or processing route changes, limiting transferability and slowing discovery. In this project, we will develop Recurrent Neural Operator++ (RNO++), a foundation model for constitutive behaviour that unifies materials-knowledge representations with operator learning. The project is fully funded at the UK home rate and will be hosted within the UKRI AI CDT. It is funded by the UK Atomic Energy Authority and will be based at the University of Cambridge, with an expected start date of September 2026.