Building a large-scale physics and ML-based automated insurance quoting system that uses property-specific data to incentivice mitigation strategies. Stack includes python, typescript, Go, Kubernetes, docker.

Development of a high-performance CFD (Computational Fluid Dynamics) analysis and optimization cloud platform. Responsible for the initial version of the GPU solver written in C++ and Cuda. Additional tasks include the implementation of the communication protocol of the solver component with the backend (go/grpc/protobuf), post-processing of the solution data (C++/threading) and frontend development (Typescript/React).

Coordination of voluntary work in a non-profit organization

(https://su2foundation.org) focused on research and education in engineering sciences. Part of the organizing team for the annual SU2 conference.

Management, maintenance and general code development of the open-source CFD code SU2. Major contributions included a replacement of the build system (meson and ninja) and the addition of automated testing (Travis CI) and distribution (docker).

Extended the open-source CFD software package SU2

(https://su2code.github.io) to be able to compute gradients in an automatic fashion using Algorithmic Differentiation on HPC platforms. This involved the application of advanced C++17 templating features as well as modifications to the existing distributed parallel communication layer (MPI) to enable the reversal of the computational graph.