Denghui Lu
Postdoctoral Researcher at ETH Zürich
Zurich, Zurich, Switzerland
Summary
🤩
Rockstar🎓
Top SchoolDenghui Lu is a postdoctoral researcher and software engineer based in Zurich with six years of experience building high-performance scientific computing and ML-enabled simulation tools. His work blends backend systems, GPU engineering, and numerical correctness—contributing to notable open-source projects like DeepMD-kit (NNP interfaces and device-memory fixes) and ABACUS (multi-device and ROCm GPU support). Trained at Peking University (PhD) and Sun Yat-sen University (BE), he focuses on optimizing many-body potential representations and electronic-structure kernels for production-scale workloads. Colleagues value his attention to numerical consistency, cross-device support, and pragmatic fixes that prevent subtle runtime bugs.
6 years of coding experience
Doctor of Philosophy - PhD, Doctor of Philosophy - PhD at Peking UniversityBachelor of Engineering - BE, Computer Science, Bachelor of Engineering - BE, Computer Science at Sun Yat-sen University
Github Skills (17)
c-language10
molecular-simulation10
deep-learning10
molecular-dynamics-simulation10
nonlinear-dynamics10
cuda10
cprogramming-language10
linear-algebra10
cluster-computing9
machine-learning9
parallel-computing9
tensorflow9
performance-optimization9
scientific-computing9
unit-test8
Programming languages (2)
Github contributions (5)
deepmodeling/deepmd-kit
Sep 2019 - Sep 2022
A deep learning package for many-body potential energy representation and molecular dynamics
Role in this project:
Back-end Developer & ML Engineer
Back-end Developer & ML EngineerContributions:151 reviews, 210 commits, 109 PRs in 3 years
Contributions summary:Denghui primarily contributed to the code related to potential energy representation and molecular dynamics. Their commits include modifications to the C++ code for the Neural Network Potential (NNP) interface, including changes to the data structure and variable definitions, and fix of a potential bug of accessing device memory. These changes indicate a focus on optimizing the performance and correctness of the NNP model within the DeepMD-Kit framework. Furthermore, the user demonstrated familiarity with the integration and optimization of the model.
lammpspythonipideepmdtensorflow
abacusmodeling/abacus-develop
Jul 2021 - Jan 2023
An electronic structure package based on either plane wave basis or numerical atomic orbitals.
Role in this project:Back-end Developer & GPU Engineer
Back-end Developer & GPU EngineerContributions:43 commits, 1 comment in 1 year 6 months
Contributions summary:Denghui primarily focused on adding and optimizing GPU support for various components within the ABACUS electronic structure package. Contributions include the addition of multi-device support for the ekinetic hpsi and psiToRho functions, as well as GPU support for the CG method. They addressed compilation errors, memory leaks, and CI failures while ensuring numerical consistency. Moreover, they added device templates for several classes and implemented ROCm support.
chemical-engineeringelectronic-structureplane-waveorbitalspackage-based
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.