Mark Hoemmen - Architect at NVIDIA

Mark Hoemmen

Architect at NVIDIA

Albuquerque, New Mexico, United States

Join Prog.AI to see contacts

Summary

🤩

Rockstar

🎓

Top School

Mark Hoemmen is an architect at NVIDIA with 15 years of experience designing high-performance parallel software for CPUs, GPUs, and heterogeneous systems. He specializes in parallel algorithms, numerical linear algebra, and performance portability, contributing to C++ Standard proposals and production-grade libraries like Kokkos and RAPIDS' raft. His work spans low-level optimization, fault-tolerant algorithms, and automatic performance tuning across MPI, OpenMP, CUDA, and modern C++ features such as mdspan. At Sandia and in industry roles he bridged research and engineering, shipping robust scientific software and fixing tricky memory/atomic bugs that improved stability and performance. Mark’s contributions to widely used open-source projects include modernizing interfaces (mdspan-ifying core ML primitives) and optimizing scratch-memory and sort paths, reflecting a pragmatic mix of research rigor and production-savvy. He holds a PhD in Computer Science from UC Berkeley and is based in Albuquerque, NM.

15 years of coding experience

13 years of employment as a software developer

PhD Computer Science, PhD Computer Science at University of California, Berkeley

Bachelor of Science - BS Mathematics and Computer Science, Bachelor of Science - BS Mathematics and Computer Science at University of Illinois Urbana-Champaign

English, German

Github Skills (22)

c-language10

machine-learning10

cluster-computing10

parallel-computing10

gpu10

cuda10

cprogramming-language10

linear-algebra10

scientific-computing10

k10

debug9

debugging9

building-blocks9

abstraction9

data-structures8

Programming languages (13)

C++CSSCRustCMakeTeXMakefilePerl

Github contributions (5)

rapidsai/raft

Sep 2022 - Oct 2022

RAFT contains fundamental widely-used algorithms and primitives for machine learning and information retrieval. The algorithms are CUDA-accelerated and form building blocks for more easily writing high performance applications.

Role in this project:

Back-end Developer & ML Engineer

Contributions:110 reviews, 9 commits, 11 PRs in 29 days

Contributions summary:Mark contributed to the `raft` library by fixing memory management issues related to `malloc` and `delete[]` mismatches, improving code stability. They also modernized the codebase by adding `mdspan` overloads for several functions, including `make_regression`, `permute`, `sampleWithoutReplacement`, and `rmat_rectangular_gen`, enhancing the library's usability. Additionally, the user developed a new `mdspan`-ified `multi_variable_gaussian` interface, which is a core component for machine learning projects.

sciencedata-sciencemachine-learninggraph-data-sciencefundamental

kokkos/kokkos

May 2015 - Apr 2019

Kokkos C++ Performance Portability Programming Ecosystem: The Programming Model - Parallel Execution and Memory Abstraction

Role in this project:

Back-end Developer

Contributions:12 reviews, 22 commits, 15 PRs in 4 years

Contributions summary:Mark primarily contributed to bug fixes and enhancements within the Kokkos library, addressing issues related to complex number operations, atomic operations, and memory management. They improved the performance of the scratch memory space by conditionally enabling debugging output. Furthermore, they refactored and improved the library's resize functionality and incorporated optimizations by leveraging GNU parallel sort. These changes indicate a focus on improving the core functionality, stability, and performance of the Kokkos library.

memorympic-plus-plusmulti-threadingkokkos

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.

Request Free Trial