Marc De Frahan is a Senior Developer Technology Engineer specializing in GPU-enabled and AI-powered numerical simulations for energy and high-energy-density applications, now at NVIDIA after a decade in computational science. He combines deep expertise in C/C++ GPU programming (CUDA, HIP, Kokkos), Python ML (TensorFlow, PyTorch), and high-order numerical methods like Discontinuous Galerkin to accelerate multi-GPU exascale workflows. His Ph.D. work at the University of Michigan applied supercomputing to fluid and wave interactions across domains from supernovae to combustion and biomedical therapies, reflecting a rare breadth between fundamental physics and production-grade software. At the National Laboratory of the Rockies he moved research prototypes toward scalable HPC implementations and contributed practical packaging support to spack/spack, expanding access to community simulation tools like amr-wind and openfast. He enjoys science outreach and translating complex ideas for diverse audiences, taking particular delight in sparking that “aha” moment in learners. Based in Golden, Colorado, Marc blends rigorous applied mathematics training with hands-on GPU and ML engineering to drive performant, real-world simulation systems.
10 years of coding experience
9 years of employment as a software developer
Ph.D Mechanical Engineering, Ph.D Mechanical Engineering at University of Michigan
M.S Applied Mathematics Engineering, M.S Applied Mathematics Engineering at Université catholique de Louvain
École polytechnique de Louvain - Louvain School of engineering
University of California, Los Angeles
Applied Mathematics, Applied Mathematics at INSA Toulouse - Institut National des Sciences Appliquées de Toulouse
A flexible package manager that supports multiple versions, configurations, platforms, and compilers.
Role in this project:
Backend Developer
Contributions:4 reviews, 6 PRs, 11 comments in 10 months
Contributions summary:Marc primarily contributed to the `spack/spack` repository by adding and updating package definitions for scientific computing software. Their work included adding versions of `amr-wind` and `openfast`, integrating a `rosco` package, and introducing variants for features like FPE trapping and fast farm capabilities. These modifications demonstrate a focus on expanding the repository's support for tools used in high-performance computing and scientific simulation.
Lattice Boltzmann method solver using adaptive mesh refinement
Contributions:1 review, 75 PRs, 86 pushes in 1 year 5 months
adaptive-mesh-refinementlattice-boltzmann-method
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.