江振瑜@googs1025

全文搜索

输入关键词开始搜索

    江振瑜

    About

    CYJiang

    CYJiang · @googs1025

    1. About Me

    I’m Jiang Zhenyu (CYJiang). My open source work focuses on Kubernetes scheduling, GPU infrastructure, LLM inference infrastructure, and accelerator-aware workload orchestration.

    In my current role, I help grow the MUSA open source ecosystem and explore its integration with cloud native and AI infrastructure.

    Previously, I worked on cloud platform engineering at ByteDance.

    I am a slow walker, but I never walk backwards.

    2. Open Source Focus

    Kubernetes Scheduling

    Scheduling, resource orchestration, batch workloads, and related projects including scheduler-plugins, descheduler, Volcano, and Koordinator.

    LLM Inference Infrastructure

    Cloud native infrastructure for scalable model serving, including Aibrix, llmaz, vLLM, SGLang, and llm-d.

    GPU & Heterogeneous Computing

    GPU management, Kubernetes device integration, accelerator-aware scheduling, CUDA, and MUSA.

    3. Open Source Roles & Recognition

    Kubernetes Member

    Contributing to Kubernetes and the SIG Scheduling ecosystem.

    Aibrix Maintainer

    Supporting the development and community of cloud native GenAI inference infrastructure.

    Kubernetes Contributor Award 2025 — SIG Scheduling

    Recognized for contributions across Kubernetes scheduler, descheduler, and scheduler-plugins.

    Official recognition

    Reviewer

    Participating in design discussions and code reviews for scheduler-plugins, descheduler, and llmaz.

    Community Member

    Participating in the Volcano and Koordinator communities.

    4. MUSA Ecosystem

    fake-gpu-musa

    Simulates MUSA and CUDA GPU components in Kubernetes environments, including user-space libraries, command-line tools, and device integration.

    musa-learning-notes

    Notes and experiments from learning the MUSA SDK and GPU computing stack.

    5. Currently Exploring

    • MUSA and GPU computing fundamentals
    • GPU management, device integration, and scheduling on Kubernetes
    • Large-scale LLM inference workloads

    6. Contact

    7. 中文简介

    我是江振瑜（CYJiang）。我的开源工作主要关注 Kubernetes 调度、GPU 基础设施、LLM 推理基础设施和异构算力工作负载编排。目前，我也会在工作中协助建设和扩展 MUSA 开源生态，探索 MUSA 与云原生及 AI 基础设施的结合。此前曾在字节跳动从事云平台相关工作。

    我是 Kubernetes Member、Aibrix Maintainer，以及 scheduler-plugins、descheduler 和 llmaz Reviewer，同时参与 Volcano 与 Koordinator 社区。2025 年，我获得了 Kubernetes SIG Scheduling Contributor Award。

    当前主要学习和探索 MUSA、GPU 计算、Kubernetes GPU 管理与调度，以及大规模 LLM 推理工作负载。