Zhanghao Wu
Something New at Something new
Berkeley, California, United States
Summary
🤩
Rockstar🎓
Top SchoolZhanghao Wu is a systems-oriented PhD-trained software engineer based in Berkeley with nine years of experience building efficient ML infrastructure and cloud-native tooling. At UC Berkeley's Sky Computing Lab he led creation of SkyPilot—an open-source platform for running AI workloads across Kubernetes and 15+ clouds—and earned a Best Paper at NSDI 2024 for systems work that bridges research and production. His background includes efficient deep learning hardware co-design at MIT HAN Lab and applied sequence modeling at ByteDance, giving him a rare blend of ML research and backend/DevOps chops. He focuses on reliability and cost-efficient execution (automatic API provisioning, retry logic, resource refactors) and often contributes both code and practical cloud docs to make complex systems usable.
9 years of coding experience
2 years of employment as a software developer
Doctor of Philosophy - PhD Computer Science, Doctor of Philosophy - PhD Computer Science at University of California, BerkeleyResearch Assistant at HAN LAB MIT Computer Science, Research Assistant at HAN LAB MIT Computer Science at Massachusetts Institute of TechnologyBachelor's degree Computer Science ACM Honors Class, Bachelor's degree Computer Science ACM Honors Class at Shanghai Jiao Tong University
Github Skills (7)
kubernetes10
python10
kubernetes-pods10
gcp9
aws9
terraformer3
terraform3
Programming languages (7)
Github contributions (5)
skypilot-org/skypilot
Nov 2021 - Jan 2023
SkyPilot: Run AI and batch jobs on any infra (Kubernetes or 14+ clouds). Get unified execution, cost savings, and high GPU availability via a simple interface.
Role in this project:
Backend & DevOps Engineer
Backend & DevOps EngineerContributions:6 releases, 4099 reviews, 626 commits in 1 year 3 months
Contributions summary:Zhanghao contributed to both backend and infrastructure aspects of the SkyPilot project. They added documentation for cloud access (AWS permissions), implemented new features like automatic provisioning of the API server, and refactored and improved several existing features, notably related to the cluster name and user identity. The user also worked on performance improvements by refactoring the code to avoid unnecessary resource usage. They added retry logic for various operations to improve the robustness of the platform.
pythonspot-instancesserverlesstpuml-infrastructure
Michaelvll/ray
Apr 2020 - Jul 2023
A fast and simple framework for building and running distributed applications. Ray is packaged with RLlib, a scalable reinforcement learning library, and Tune, a scalable hyperparameter tuning library.
Contributions:170 pushes, 4 branches in 3 years 3 months
scalableraydistributed-applicationshyperparametersimple-framework
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.