Ning Wang - Software Engineer at Meta

Ning Wang

Software Engineer at Meta

Menlo Park, California, United States

Join Prog.AI to see contacts

Summary

🤩

Rockstar

🎓

Top School

Ning Wang is a software engineer with a decade of experience building large-scale machine learning and computer vision systems, currently working at Meta Superintelligence Labs after a stint focused on efficient GenAI model training at Databricks. He has deep expertise in distributed PyTorch model parallelism and recommendation-system infrastructure, contributing notable fixes and performance work to high-profile open-source projects like torchrec and mosaicml/composer. Ning’s background spans production computer vision at Baidu to optimizing inference and sharding strategies, and he routinely improves robustness in training (e.g., resumable profilers, FSDP resharding and checkpoint retry mechanisms). Based in Menlo Park, he blends research-caliber ML knowledge with pragmatic backend engineering and a track record of shipping maintainable, high-performance code across industry-scale systems.

11 years of coding experience

10 years of employment as a software developer

Bachelor of Science (BS) Mathematics and Computer Science, Bachelor of Science (BS) Mathematics and Computer Science at Northeastern University, China

Master’s Degree Computer Science, Master’s Degree Computer Science at New York University - Polytechnic School of Engineering

English, Chinese

Github Skills (15)

pytorch10

machine-learning10

deeplearning-ai10

distributed-systems10

deep-learning10

recommendation-system10

python10

sharding10

ml10

cprogramming-language9

parallelization9

sdp9

c-language9

unit-testing9

cuda7

Programming languages (2)

C++Python

Github contributions (5)

mosaicml/composer

Feb 2024 - Feb 2025

Supercharge Your Model Training

Role in this project:

ML Engineer

Contributions:3 releases, 119 reviews, 88 PRs in 1 year

Contributions summary:Ning primarily contributed to the composer/profiler and torch_profiler modules, addressing issues related to profiling and resuming training runs. They implemented fixes to accurately handle the skip_first parameter within the cyclic schedule during resumption. They also moved an after_load callback to the profiler, along with fixing unit tests. Further contributions included FSDP resharding after an OOM condition and the addition of retry mechanisms for checkpoint downloads.

pytorchml-systemsdeep-learningneural-networksmachine-learning

meta-pytorch/torchrec

Nov 2021 - Aug 2022

Pytorch domain library for recommendation systems

Role in this project:

Back-end Developer & ML Engineer

Contributions:3 reviews, 14 commits, 14 PRs in 9 months

Contributions summary:Ning contributed to the PyTorch domain library for recommendation systems by implementing and improving core functionalities related to model performance and distributed training. They focused on optimizing inference performance by migrating code to C++ and improving the KeyedJaggedTensor (KJT) operations. Furthermore, the user made improvements to the embedding tower sharding and partitioner, supporting distributed model training and deployment. Additional contributions included fixing unit tests and adding documentation, showcasing a focus on code quality and maintainability.

pytorchrecommendation-systemgpudeep-learningcuda

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.

Request Free Trial