Solutions Architect for High-Performance Computing in the Cloud
Director of Cambridge Supercomputing
I led the Research Software Engineering (RSE) team in the Research Computing Services Division at the University of Cambridge.
The team works on a variety of technical projects in various fields, from digital humanities to astrophysics, with a focus on porting and optimisation of HPC applications. The team works with researchers at the University of Cambridge as well as other national and international collaborations, pursuing the objective "Better Software for Better Research". We also work closely with vendors and industry partners to understand how scientific applications will perform on next-generation computer hardware.
As a Research Software Engineer and High-Performance Computing Consultant, I optimised the performance of complex scientific codes across a range of projects. This work included porting and optimisation of HPC applications, from a wide range of scientific domains, enabling scientists to make use of the cutting-edge heterogeneous hardware at the forefront of scientific computing.
My team contributed to the ExoAI collaboration, developing machine learning methods for exoplanet discovery.
(With Krishna Kumar) I contributed to a novel, scalable implementation of the Material Point Method (MPM)
(With Andy Turner) I contributed to a report comparing the performance of some key applications on the various HPC systems available to researchers in the UK.
Report on performance and scalability of OpenFOAM and Cloverleaf, two CFD applications, on the ARM platform.
Also on HPC Wire.
The ignition of condensed-phase explosives is investigated by simulating the reflection of a shock through nitromethane. The passage of the shock leads to the formation of a reaction wave. Of particular interest is the evolution of the reaction wave from when it emerges to when it eventually overtakes the precursor shock. We identify, locate and characterise transient, quasi-steady phenomena in the unsteady flow prior to the formation of a steady detonation wave. Results indicate that the reaction wave is composed of transient, quasi-steady waves similar to the ones observed in the gaseous case, albeit in a more complex combination.