Mengxuan Li
系统架构师 at 上海密瓜智能科技有限公司
Beijing, China
Summary
🤩
Rockstar🎓
Top SchoolMengxuan Li is a systems architect based in Beijing with six years of experience building cloud-native and AI infrastructure, currently leading architecture at 上海密瓜智能科技. He is the founder and maintainer of Project-HAMi, an open-source heterogeneous AI computing virtualization middleware, and a maintainer for the CNCF-incubating Volcano project, demonstrating deep expertise in GPU sharing and Kubernetes scheduling. Previously an architect at 4Paradigm, he focuses on scheduler correctness and device-aware resource allocation, having contributed targeted fixes to scoring logic in HAMi’s scheduler. A CNCF speaker with a Tsinghua computer science master’s background, he combines research-grade rigor with hands-on operational improvements that reduce resource fragmentation in AI clusters. Not obvious from titles alone: his contributions frequently bridge scheduler math and practical device requirements, improving real-world placement accuracy for heterogeneous accelerators.
6 years of coding experience
硕士, 计算机, 硕士, 计算机 at 清华
Github Skills (8)
scheduling10
algorithms10
go10
sched10
scheduler10
kubernetes-pods9
device-management9
kubernetes9
Programming languages (7)
Github contributions (5)
Project-HAMi/HAMi
Aug 2021 - Jan 2023
Heterogeneous AI Computing Virtualization Middleware
Role in this project:
Back-end Developer
Back-end DeveloperContributions:15 reviews, 262 commits, 216 PRs in 1 year 4 months
Contributions summary:Mengxuan primarily focused on fixing issues within the scheduler's scoring mechanism. They addressed problems in the `calcScore` function related to calculating scores for nodes and handling device resource requests. The commits involve code changes in the `pkg/scheduler/score.go` file, indicating a focus on improving the accuracy and efficiency of the scheduling process within the project. The changes also included adding new fields to support the device requirements.
memorygpuvirtualizeschedulerdevice-plugin
4paradigm/k8s-vgpu-scheduler
Mar 2021 - Jul 2022
OpenAIOS vGPU device plugin for Kubernetes is originated from the OpenAIOS project to virtualize GPU device memory, in order to allow applications to access larger memory space than its physical capacity. It is designed for ease of use of extended device memory for AI workloads.
Contributions:58 commits, 5 PRs, 68 pushes in 1 year 3 months
memoryextendedgpuvirtualizeworkloads
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.