Wei Wu
Senior Software Engineer at NVIDIA
Los Alamos, New Mexico, United States
Summary
🤩
Rockstar🎓
Top SchoolWei Wu is a Senior Software Engineer with 12 years of experience specializing in high-performance computing, distributed and parallel programming models, and runtime systems for large-scale heterogeneous supercomputers. He has contributed to prominent projects like Legion and Open MPI and focused on memory layout optimizations and HIP integration to boost performance on modern accelerators. At NVIDIA and previously Los Alamos National Laboratory he built and optimized task-based runtimes for scientific and ML workloads, often hunting down scalability and performance bottlenecks in distributed deep learning. His GitHub work on FlexFlow shows hands-on backend and ML engineering skills—improving interoperability with Keras, ONNX, and PyTorch while debugging core C++ runtime issues. Combining deep research training (PhD) with production-grade engineering, he bridges academic systems research and practical implementation. Based in Los Alamos, he blends low-level optimization expertise with a practical eye for making runtimes robust and model-compatible at scale.
12 years of coding experience
11 years of employment as a software developer
Doctor of Philosophy (Ph.D.), Computer Science, Doctor of Philosophy (Ph.D.), Computer Science at University of Tennessee-KnoxvilleMaster of Science (M.S.), Computer Engineering, Master of Science (M.S.), Computer Engineering at Purdue UniversityBachelor of Engineering (B.Eng.), Computer Software Engineering, Bachelor of Engineering (B.Eng.), Computer Software Engineering at Beijing Institute of Technology
Github Skills (18)
c-language10
memory-management10
data-structure10
keras10
data-structures10
cprogramming-language10
parallelization9
parallel9
deep-q-learning9
parallel-computing9
performance-optimization9
deep-learning9
parallel-processing9
cuda9
machine-learning8
Programming languages (8)
Github contributions (5)
flexflow/flexflow-train
Feb 2020 - Jan 2023
Automatically Discovering Fast Parallelization Strategies for Distributed Deep Neural Network Training
Role in this project:
Back-end Developer & ML Engineer
Back-end Developer & ML EngineerContributions:1 release, 25 reviews, 697 commits in 2 years 11 months
Contributions summary:Wei's commits primarily focus on debugging and refining existing code within the FlexFlow framework. The changes suggest a focus on improving the core functionality of the library, as evidenced by the fixes to print statements in the Keras implementation and addressing specific issues within the C++ backend (e.g., issues with use of bias). The user demonstrates a familiarity with the library's internal structure and codebase, as well as a working knowledge of Keras API. The modifications to the ONNX and PyTorch examples, the creation of new ones, and changes for the model interface suggests the user is actively involved in ensuring compatibility and smooth operation of the framework with different deep learning models.
pytorchflexibledeep-learningmachine-learningdistributed-deep-learning
StanfordLegion/legion
Jul 2017 - Jan 2023
The Legion Parallel Programming System
Role in this project:Back-end Developer
Back-end DeveloperContributions:281 commits, 1 PR, 339 comments in 5 years 7 months
Contributions summary:Wei's commits focus on adding and modifying functionalities related to attaching and manipulating array data within the Legion Parallel Programming System. The changes include implementing mechanisms for attaching single and multiple fields to physical regions, specifically for SOA (Structure of Arrays) and AOS (Array of Structures) layouts, which demonstrates a focus on memory management and data structure optimization. The user also worked on refining the build process related to HIP (Heterogeneous Interface for Portability) and the addition of features such as enabling the task through put for HIP, indicating efforts to improve performance and integration with HIP. These contributions suggest a focus on enhancing the core capabilities and performance of the Legion system.
system-programminglegionparallel-programmingparallel
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.