Runyu Lu
Research Intern at NVIDIA
Ann Arbor, Michigan, United States
Summary
🤩
Rockstar🎓
Top SchoolRunyu Lu is a PhD student in Computer Science at the University of Michigan and a research intern at NVIDIA with five years of experience in robotics and ML systems. He brings hands-on expertise in high-performance neural network inference—contributing x86 and AVX512 optimizations, SIMD implementations, and GridSample support to the widely used Tencent/ncnn mobile inference framework. Trained with a top GPA from Huazhong University of Science and Technology, Runyu focuses on bridging research and production performance, particularly on CPU-level optimizations for common deep learning layers. He combines academic rigor with practical engineering, routinely turning low-level algorithmic improvements into measurable throughput gains for real-world ML workloads.
5 years of coding experience
Doctor of Philosophy - PhD, Computer Science, Doctor of Philosophy - PhD, Computer Science at University of MichiganBachelor's degree, Computer Science, 3.95, Bachelor's degree, Computer Science, 3.95 at Huazhong University of Science and Technology
Github Skills (17)
vulkan10
c-language10
inference10
x8610
deeplearning-ai10
deep-learning10
neon10
neural-network10
simd10
cprogramming-language10
avx9
relu9
tensorflow8
onnx8
pytorch8
Programming languages (6)
Github contributions (5)
Tencent/ncnn
Apr 2022 - Nov 2022
ncnn is a high-performance neural network inference framework optimized for the mobile platform
Role in this project:
ML Engineer
ML EngineerContributions:12 reviews, 8 commits, 68 PRs in 6 months
Contributions summary:Runyu primarily contributed to optimizing and extending the ncnn framework, focusing on x86 CPU optimizations. Their work included implementing AVX512 intrinsics for BatchNorm, merging multiple elempack functionalities, and integrating intrinsics for ELU and PReLU activation functions. The user also contributed to the GridSample operation, adding support for different sampling types. This involved optimizing performance with SIMD instructions for common neural network layers.
simdncnnvulkaniostensorflow
LRY89757/ColossalAI
Mar 2024 - Oct 2024
Making large AI models cheaper, faster and more accessible
Contributions:94 pushes, 13 branches in 7 months
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.