Matthew Deng
Software Engineer at Anyscale
San Francisco Bay Area United States
Summary
🤩
Rockstar🎓
Top SchoolMatthew Deng is a software engineer with a decade of experience building scalable systems from his UC Berkeley CS roots to core engineering roles at LinkedIn and now Anyscale in the San Francisco Bay Area. He focuses on backend and distributed systems work, with recent contributions to Ray—an influential open-source AI compute engine—where he improved distributed PyTorch training, GPU stream support, and serialization for Ray Tune. At LinkedIn he progressed from entry-level engineer to staff, demonstrating both technical depth and the ability to influence large production systems. His profile blends hands-on MLOps and backend engineering, making him fluent in the intersection of ML infrastructure and production-grade software. Notably, he balances deep implementation work (core training loop refactors and test-driven fixes) with long-term platform thinking at a fast-growing AI infrastructure company.
10 years of coding experience
5 years of employment as a software developer
Bachelor of Science (BS), Electrical Engineering and Computer Science, Bachelor of Science (BS), Electrical Engineering and Computer Science at University of California, BerkeleyLynbrook High School
Github Skills (6)
gpu-programming10
pytorch10
machine-learning10
distributed-training10
python10
testing9
Programming languages (4)
Github contributions (5)
ray-project/ray
Jun 2021 - Dec 2022
Ray is an AI compute engine. Ray consists of a core distributed runtime and a set of AI Libraries for accelerating ML workloads.
Role in this project:
Back-end Developer & MLOps Engineer
Back-end Developer & MLOps EngineerContributions:1 release, 2122 reviews, 119 commits in 1 year 6 months
Contributions summary:Matthew's contributions are primarily focused on refactoring the training pipeline for Ray Tune, specifically within the context of distributed deep learning. They integrated improvements to the core training loop functions, adding support for the `torch.cuda.Stream` and also added test cases to validate the fixes implemented. This included enhancing the internal workings of the PyTorch execution environment to facilitate GPU-based training, with improvements to serialization.
pythonconsistsruntimetensorflowserving
matthewdeng/ray
May 2021 - Mar 2025
An open source framework that provides a simple, universal API for building distributed applications. Ray is packaged with RLlib, a scalable reinforcement learning library, and Tune, a scalable hyperparameter tuning library.
Contributions:3 reviews, 1 PR, 623 pushes in 3 years 10 months
apirayscalabledistributed-applicationshyperparameter
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.