I am part of the Statistical Learning team at Zalando SE. Here, I add GPU computing expertise to a team that investigates state of the art machine learning solutions to problems and challenges faced by Europe’s number one fashion e-tailer.
Previously, I was a senior developer for electromagnetic solvers at Altair Engineering (following a merger with EM Software & Systems). This entails working on various aspects of the computational kernel of the electromagnetic modelling and simulation package FEKO, now part of the HyperWorks suite of computer aided engineering software.
Prior to that, I was a member of the computational electromagnetics group (CEMAGG) at the electrical and electronic engineering deparment of Stellenbosch University from January 2006 to July 2011. As part of the group, I undertook research in the field of computational electromagnetics (CEM) as well as general purpose GPU computing.
My PhD dissertation considered the application of GPU acceleration to matrix-based methods in CEM, and more specifically considered the use of CUDA from NVIDIA to improve the performance of the finite element method (FEM) and the method of moments (MoM) — also called the boundary element method. In the case of the MoM, the implementations successfully overcome the limitations imposed by the amount of memory available on a GPU.