Sheng Zha
Senior Applied Science Manager at Amazon
San Jose, California, United States
Summary
🤩
Rockstar🎓
Top SchoolSheng Zha is a Senior Applied Science Manager based in San Jose with a decade of experience building and leading large-scale foundation model training and acceleration efforts at Amazon and AWS. He combines hands-on systems and ML engineering—contributions to projects like Apache MXNet, Gluon-NLP, and mshadow show deep expertise in build systems, distributed training, and CUDA/C++ libraries—with product-focused applied science that powered services such as CodeWhisperer, Lex, Comprehend, and Kendra. As a founding member of the Python Data API Standards consortium and former Apache MXNet PMC chair, he blends open-source stewardship with enterprise delivery. Sheng’s background spans scalable ML platforms, RLHF and LLM pre-training, and low-level performance work such as OpenBLAS and MKL integration, reflecting both research and production chops. Colleagues describe him as a manager who still dives into build scripts and model fine-tuning, a habit that keeps teams technically grounded. He holds advanced CS training from the University of Maryland and an appetite for pragmatic, cross-layer optimizations that accelerate foundation models.
10 years of coding experience
11 years of employment as a software developer
Non-degree Computer Science, Non-degree Computer Science at The University of New MexicoMaster's degree Computer Science, Master's degree Computer Science at University of MarylandBachelor of Science - BS Computer Science, Bachelor of Science - BS Computer Science at Shanghai Jiao Tong University
Italian, Japanese, English, Chinese
Stackoverflow
629reputation
27kreached
5answers
0questions
cuda
top-5%
Github Skills (50)
dependency-management10
c-language10
python10
scripting10
mxnet10
machine-learning10
build-system10
gluon10
cmake10
dm10
openmp10
script10
deep-learning10
build-automation10
fine-tuning10
Programming languages (14)
Github contributions (5)
apache/mxnet
Jan 2017 - Mar 2022
Lightweight, Portable, Flexible Distributed/Mobile Deep Learning with Dynamic, Mutation-aware Dataflow Dep Scheduler; for Python, R, Julia, Scala, Go, Javascript and more
Role in this project:
Backend Engineer & Build Engineer
Backend Engineer & Build EngineerContributions:4 releases, 270 reviews, 305 commits in 5 years 2 months
Contributions summary:Sheng's primary contribution centers around the build process and dependency management for the MXNet deep learning framework. They developed and modified scripts to build OpenBLAS, manage PS-Lite dependencies, configure S3 access, and build image libraries. Furthermore, they created scripts for creating the build and wheel packages, indicating a focus on ensuring the framework's compatibility and ease of use in different environments. The user also worked on updating build scripts and ensuring compatibility with MacOS.
pythonschedulerdataflowmutationdata-science
dmlc/gluon-nlp
Apr 2018 - Apr 2021
NLP made easy
Role in this project:Back-end Developer
Back-end DeveloperContributions:4 releases, 90 reviews, 154 commits in 3 years 1 month
Contributions summary:Sheng primarily focused on code improvements related to the `gluon-nlp/data` and `gluonnlp/model` modules, addressing code style and bug fixes. The contributions include updates to the linting process, modifications to sampling methods, and fixes for vocabulary serialization issues, demonstrating involvement in data processing and model architecture. The user also participated in documentation updates, indicating a focus on code quality and maintainability.
nlpmxnetnatural-language-understandingnlgnatural-language-generation
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.