Minseok Lee
Engineering Manager at NVIDIA
United States, Republic Of Korea
Summary
🤩
RockstarMinseok Lee is an Engineering Manager with seven years of experience specializing in GPU-accelerated machine learning infrastructure and developer tooling. Currently managing AI DevTech at NVIDIA, he blends hands-on backend and ML engineering with people leadership to deliver high-performance recommendation systems. His open-source contributions to NVIDIA-Merlin/HugeCTR — including implementing Reshape and Concat layers, adding 3D input support to MatrixMultiply, and enabling MMoE capabilities — reflect deep expertise in productionizing CTR training frameworks. Based in the U.S. with roots in South Korea, he bridges international perspectives and large-scale GPU engineering practices. Colleagues describe him as a pragmatic problem-solver who moves efficiently between low-level performance work and cross-functional program delivery.
7 years of coding experience
Github Skills (9)
cuda10
machine-learning10
c-language10
deep-learning10
tensorflow10
cprogramming-language10
performance-optimization8
mask-rcnn7
faster-rcnn7
Programming languages (3)
Github contributions (5)
NVIDIA-Merlin/HugeCTR
Sep 2019 - Jan 2023
HugeCTR is a high efficiency GPU framework designed for Click-Through-Rate (CTR) estimating training
Role in this project:
Back-end Developer & ML Engineer
Back-end Developer & ML EngineerContributions:29 releases, 11 reviews, 559 commits in 3 years 4 months
Contributions summary:Minseok implemented and tested the Reshape layer and its associated unit tests, contributing to the core functionality of the HugeCTR framework. They added missing source files and tests for the Concat layer, as well as incorporated support for 3D inputs within the MatrixMultiply layer. This work focused on expanding the capabilities of the framework for recommendation model development, including support for multi-task models (MMoE).
cudapytorchcppgpu-accelerationdeep-learning
minseokl/HugeCTR
Oct 2020 - Mar 2023
HugeCTR is a high efficiency GPU framework designed for Click-Through-Rate (CTR) estimating training
Contributions:5 pushes in 2 years 5 months
cudapytorchgpuestimatingclick-through-rate
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.