Pak Markthub
Senior Software Engineer at NVIDIA
California, United States
Summary
🤩
Rockstar🎓
Top SchoolPak Markthub is a Senior Software Engineer at NVIDIA with nine years of experience building low-latency, high-bandwidth GPU communication technologies and earlier academic roots in a Ph.D.-level study of efficient data movement. He designs and implements GPUDirect capabilities that let GPUs talk directly to third-party devices, reducing CPU involvement and unlocking peak hardware performance for both small and large transfers. At NVIDIA he moved from postdoctoral research into production engineering, focusing on practical, deployable systems — including improving build and packaging workflows for the widely used gdrcopy library to simplify kernel module distribution via DKMS. His background spans national labs and supercomputing centers where he explored GPU memory semantics, checkpoint/restart, and extending GPU virtual memory to non-volatile storage. Based in California but with deep ties to Tokyo Institute of Technology, he blends systems research rigor with hands-on DevOps and release engineering. Colleagues can expect a developer who balances performance-first architecture with pragmatic build and deployment improvements that accelerate adoption.
9 years of coding experience
3 years of employment as a software developer
Bachelor’s Degree, Computer Engineering, 3.97, Bachelor’s Degree, Computer Engineering, 3.97 at Kasetsart UniversityDoctor of Science, Mathematical and Computing Sciences, Doctor of Science, Mathematical and Computing Sciences at Tokyo Institute of TechnologyHigh School, Science, 3.93, High School, Science, 3.93 at TriamUdom Suksa School
Thai, Japanese, English
Github Skills (11)
debian-packaging10
packaging10
kernel-module10
kernel-mode10
rpm10
package10
build-automation10
packager10
linux9
nvidia8
lib7
Programming languages (4)
Github contributions (5)
NVIDIA/gdrcopy
Oct 2019 - Oct 2022
Role in this project:
DevOps Engineer & Build/Release Engineer
DevOps Engineer & Build/Release EngineerContributions:12 releases, 77 reviews, 90 commits in 3 years 1 month
Contributions summary:Pak primarily focused on improving the build and packaging processes for the `gdrcopy` library. Their contributions involved modifying build scripts for Debian and RPM packages, including versioning, dependency checks, and build configurations. They also updated the packaging to use DKMS for kernel module builds and cleaned up the build process, streamlining the generation of installable packages.
cudamemorynvidiagpulinux
pakmarkthub/dragon
Aug 2018 - Sep 2020
A host-based framework that transparently extends the GPU addressable global memory space beyond the host memory using NVM-backed data pointers
Contributions:22 commits, 1 PR, 49 pushes in 2 years 1 month
cudamemorygpu-programmingaddressablegpu-acceleration
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.