Jonathan Madsen
Principal Member Of Technical Staff at AMD
Austin, Texas, United States
Summary
🤩
Rockstar🎓
Top SchoolJonathan Madsen is a Principal Member of Technical Staff leading GPU profiling for AMD’s ROCm stack, bringing nine years of expertise at the intersection of high-performance computing, C++/Python engineering, and heterogeneous performance analysis. His background in radiation transport and a PhD in Nuclear Engineering inform a pragmatic approach to performance optimization, evidenced by work at NERSC and contributions to Geant4 and Monte Carlo transport tools. He’s a hands-on open-source contributor who has improved GPU-accelerated tomographic reconstruction (TomoPy), enhanced Kokkos CUDA build tooling, and tightened build and packaging workflows across projects like LIKWID and Spack. Known for tackling deep build-system and compiler-integration challenges, he combines low-level systems savvy with a focus on performance portability across CPU/GPU stacks. Based in Austin, he blends research-grade rigor with product-focused delivery, often surfacing subtle build and testing fixes that disproportionately improve downstream developer productivity.
9 years of coding experience
7 years of employment as a software developer
Doctor of Philosophy - PhD Nuclear Engineering, Doctor of Philosophy - PhD Nuclear Engineering at Texas A&M UniversityHigh School Diploma, High School Diploma at Strake Jesuit College Preparatory
Github Skills (31)
image-reconstruction10
benchmark10
c-language10
tomography10
python10
toolbox10
cluster-computing10
benchmarking10
instrumentation10
cmake10
c1110
recovery10
cicd10
c1710
performance-analysis10
Programming languages (12)
Github contributions (5)
tomopy/tomopy
Nov 2018 - Oct 2021
Tomographic Reconstruction in Python
Role in this project:
Backend & DevOps Engineer
Backend & DevOps EngineerContributions:23 reviews, 11 commits, 9 PRs in 2 years 11 months
Contributions summary:Jonathan's commits primarily focused on optimizing the build, testing, and coverage processes within the TomoPy repository. This included integrating PyCTest and configuring coverage tools (GCov), along with simplifying and fixing aspects of the build system. They also implemented improvements to CUDA-based reconstruction algorithms, enhancing performance and adding features. The user’s contributions significantly impacted the project's build processes and GPU-accelerated reconstruction capabilities.
pythontomographyreconstructiontoolbox
kokkos/kokkos
Jun 2020 - Dec 2021
Kokkos C++ Performance Portability Programming Ecosystem: The Programming Model - Parallel Execution and Memory Abstraction
Role in this project:Backend Developer
Backend DeveloperContributions:93 reviews, 43 commits, 25 PRs in 1 year 5 months
Contributions summary:Jonathan primarily focused on improving the CUDA compilation process within the Kokkos framework. Their contributions include creating systems for compiler and linker rules for `nvcc_wrapper`, which allows for using a non-clang compiler when CUDA is enabled. They also addressed compiler identification logic and added support for Clang. Furthermore, they implemented several updates related to build processes and integration.
cppparallel-executionprogramming-modelparallel-computingc-plus-plus
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.