David Wu
Developer Technology Manager at NVIDIA
Chaoyang District, Beijing, China
Summary
👤
Senior🎓
Top SchoolDavid Wu is a Developer Technology Manager at NVIDIA with 8 years in developer-facing roles and over a decade of embedded and software leadership experience. He specializes in high-performance computing and deep learning infrastructure, having made substantive contributions to NVIDIA-Merlin's HugeCTR—optimizing CUDA kernels and sparse embedding layers for production-scale CTR training. Based in Beijing, he bridges low-level GPU programming and developer enablement, guiding teams to deliver performant model training pipelines and integrating advanced communication libraries. His background in signal and information processing (Beijing Institute of Technology) and early career embedded systems work give him a strong systems-first perspective that informs both architecture and mentorship. Colleagues know him for turning complex GPU and sparse-data challenges into practical, high-throughput solutions.
8 years of coding experience
10 years of employment as a software developer
硕士, 信号与信息处理, 硕士, 信号与信息处理 at 北京理工大学
Github Skills (10)
cuda10
hashtable10
kernel10
c-language10
deep-learning10
cpp10
gpu-acceleration10
cprogramming-language10
recommender-system9
recommendation-system9
Programming languages (3)
Github contributions (5)
NVIDIA-Merlin/HugeCTR
Dec 2019 - Jun 2020
HugeCTR is a high efficiency GPU framework designed for Click-Through-Rate (CTR) estimating training
Role in this project:
Back-end Developer
Back-end DeveloperContributions:117 commits, 3 comments in 5 months
Contributions summary:David primarily contributed to the core framework of HugeCTR, specifically focusing on the embedding layer. Their commits involved modifying and refactoring CUDA kernel functions for sparse embedding, including forward and backward propagation, and weight updates. The user also worked on integrating new communication libraries and implementing features related to localized slot sparse embeddings. This indicates a focus on optimizing and extending the functionality of the high-efficiency GPU framework for CTR estimation.
cudapytorchcppgpu-accelerationdeep-learning
wl1136/testInt8Memcpy
Nov 2018 - Nov 2018
Contributions:1 branch in 1 day
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.