Jinho Suh
Principal Architect at NVIDIA
Austin, Texas, United States
Summary
🤩
Rockstar🎓
Top SchoolJinho Suh is a Principal Architect at NVIDIA with over a decade of experience shaping high-performance SoC and ML/DL inference systems, drawing on a Ph.D. in Computer Engineering from USC. He specializes in inference performance engineering, having advanced NVIDIA’s ML/DL inference stack and contributed to MLPerf inference benchmarks—optimizing 3D U-Net throughput, preprocessing, and multi-stream performance. Prior roles at ARM and Intel focused on SoC performance, on-die interconnects, cache coherence, and memory controller design, giving him rare cross-layer expertise from silicon to ML software. Based in Austin, he blends rigorous research experience (performance variability, soft-error modeling) with hands-on engineering to turn academic insights into production performance wins.
5 years of coding experience
16 years of employment as a software developer
Ph.D Computer Engineering, Ph.D Computer Engineering at University of Southern CaliforniaB.S Electrical Engineering; Control Engineering, B.S Electrical Engineering; Control Engineering at Seoul National University
English, Korean
Github Skills (8)
computer-vision10
pytorch10
machine-learning10
benchmark10
benchmarking10
perf10
mlp10
python9
Programming languages (1)
Github contributions (5)
mlcommons/inference
Jun 2021 - Nov 2022
Reference implementations of MLPerf™ inference benchmarks
Role in this project:
ML Engineer
ML EngineerContributions:18 reviews, 13 commits, 19 PRs in 1 year 5 months
Contributions summary:Jinho contributed to the development and improvement of the 3D UNet model within the MLPerf inference benchmark repository. Their work involved refactoring and bringing up the 3D UNet model for the KiTS19 kidney tumor task, including preprocessing and data handling. They also added PyTorch checkpoint models, optimized multi-stream performance, and fixed issues related to the model's integration within the broader MLPerf ecosystem. Furthermore, they made adjustments to accuracy calculations.
implementationsbenchmarkingdeep-learningmlperfinference
nv-jinhosuh/inference
May 2021 - Feb 2024
Reference implementations of MLPerf™ inference benchmarks
Contributions:97 pushes, 20 branches in 2 years 9 months
implementationsdeep-learningmlperfinferencemachine-learning
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.