GPU算力平台运维工程师（驻美国）

Job Description Location: Cupertino, CA Employment Type: Full-time Travel Requirements: On-site work at data center facilities may be required for deployment, maintenance, and troubleshooting.

Position Overview Operate and maintain high-performance GPU infrastructure centered on NVIDIA B300 systems. Ensure secure, reliable, and efficient compute operations through proactive monitoring, maintenance, incident response, and capacity management in support of ongoing compute service delivery.

Key Responsibilities 1. Perform daily health checks, configuration management, and maintenance for B300 GPU servers and associated compute, storage, and networking equipment.

• Monitor GPU/CPU usage, memory, storage, network performance, power, temperature, and system health; maintain dashboards, alerts, and logs.
• Install, configure, patch, and maintain Linux, NVIDIA drivers, CUDA environments, and supporting components.
• Diagnose hardware, GPU, network, and storage issues; manage incident severity, recovery, and root-cause analysis.
• Coordinate with colocation/data center providers, hardware vendors, and network partners on rack deployment, on-site support, and component replacement.
• Support node onboarding, cluster expansion, change management, resource monitoring, and capacity planning.
• Maintain runbooks, SOPs, inspection records, incident reports, and emergency procedures.
• Collaborate with technical teams to resolve runtime and resource issues and improve availability, mean time to recovery, and utilization.

Qualifications 1. Bachelor's degree in a relevant technical field or equivalent practical experience. 2. 3+ years administering Linux servers, data center infrastructure, cloud infrastructure, or similar systems.

• Strong knowledge of Linux, server hardware, TCP/IP networking, storage, and systems troubleshooting.
• Ability to automate routine work using Shell or Python and use monitoring, logging, and alerting tools.
• Working knowledge of Docker, Kubernetes, and GPU computing environments. 6.

Experience with incident response, change control, operational handoffs, and technical documentation.

• Availability to support scheduled maintenance and critical incident response when required.

Preferred Qualifications • Experience maintaining NVIDIA GPU systems, including B200/B300 or H100/H200 platforms.

• Familiarity with NVIDIA DCGM, NVLink/NVSwitch, InfiniBand, or RoCE.
• Experience with GPU cluster scheduling, Kubernetes GPU Operator, Slurm, or AI training/inference operations.
• Experience coordinating on-site data center operations, vendor support, or HPC systems. 工作地点：Cupertino, CA; 工作类型：Full-time 出差要求： 具体工作地点根据 GPU 设备部署机房确定，需支持现场运维 职位概述 负责以NVIDIA B300为核心的高性能GPU算力基础设施的日常运行、监控、维护与故障响应，保障计算资源的安全性、可用性和运行效率，并支持算力服务的持续交付。 岗位职责 1. 负责B300 GPU服务器及相关计算、存储、网络设备的日常巡检、配置管理和运行维护。 2. 监控GPU、CPU、内存、存储、网络、功耗、温度及系统健康指标，维护告警和日志体系。 3. 开展Linux操作系统、NVIDIA驱动、CUDA环境及相关组件的安装、升级、补丁和配置管理。 4. 排查GPU、服务器、网络和存储故障，执行事件分级、响应、恢复及根因分析。 5. 与IDC、硬件厂商、网络服务商协作，处理设备上架、现场维护、备件更换及技术支持。 6. 支持计算节点上线、集群扩容、变更管理、资源使用监控和容量规划。 7. 编写与执行运维SOP、巡检记录、故障报告、应急预案及操作文档。 8. 配合技术团队处理运行环境与资源使用问题，持续改善可用性、MTTR及资源利用效率。 任职要求 1. 计算机、网络工程、信息系统或相关专业本科及以上学历，或具备同等实践经验。 2. 3年以上Linux服务器、数据中心、云基础设施或相关平台运维经验。 3. 熟悉Linux、服务器硬件、TCP/IP网络、存储及常见系统故障诊断方法。 4. 能够使用Shell或Python完成基础运维自动化，熟悉监控、日志和告警工具。 5. 理解容器技术（Docker、Kubernetes）及GPU计算环境的基本运行方式。 6. 具有故障响应、变更控制、值班协作和技术文档编写能力。 7. 能够按业务需要参与计划内维护窗口及紧急事件响应。 加分项 • 具有NVIDIA GPU服务器（包括B200/B300、H100/H200等）维护经验。 • 熟悉NVIDIA DCGM、NVLink/NVSwitch、InfiniBand或RoCE。 • 熟悉GPU集群调度、Kubernetes GPU Operator、Slurm或AI训练/推理环境。 • 有IDC现场协作、硬件厂商故障支持或高性能计算环境经验。