【AI Infra】基础架构工程师（SRE）(J12900)

免费注册后解锁"你的匹配分 vs 高分画像"与个人技能缺口。

1、做什么：
负责 GPU 算力集群的调度、运维与优化，以及阿里云资源的维护与管理，为模型训练和数据处理提供稳定、高效、弹性的算力基础设施。

2、核心职责：
- GPU 集群（H20 等）部署、运维与资源调度系统建设
- 分布式训练环境搭建（NCCL 网络优化、RDMA 配置、Checkpoint 自动保存与恢复）
- 阿里云资源维护与混合云架构设计（弹性伸缩、成本优化）
- 集群级监控体系与告警策略，容量规划支撑算力增长
1、任职要求：
- 计算机/软件工程 等相关专业，本科及以上
- 熟悉 Linux 系统、网络（TCP/IP/RDMA）、存储原理
- 熟悉 Kubernetes，有 GPU 集群调度基础
- 了解阿里云/AWS 等公有云资源管理
- 熟悉监控告警体系（Prometheus/Grafana/ELK）

2、加分项：
- 大规模 AI 训练集群（千卡级）运维经验
- NCCL/MPI 分布式通信库与网络拓扑优化经验
- 基础设施即代码（Terraform/Pulumi）实践

1. What you will do:
Responsible for the scheduling, operations, and optimization of GPU computing clusters, as well as the maintenance and management of Alibaba Cloud resources, providing stable, efficient, and elastic computing infrastructure for model training and data processing.

2. Core responsibilities:
- Deployment, operations, and resource scheduling system development for GPU clusters (H20, etc.)
- Setup of distributed training environments (NCCL network optimization, RDMA configuration, automatic checkpoint saving and recovery)
- Alibaba Cloud resource maintenance and hybrid cloud architecture design (elastic scaling, cost optimization)
- Cluster-level monitoring systems and alerting strategies, capacity planning to support computing growth
1. Qualifications:
- Major in Computer Science/Software Engineering or related fields, bachelor's degree or above
- Familiar with Linux systems, networking (TCP/IP/RDMA), and storage principles
- Familiar with Kubernetes, with a foundation in GPU cluster scheduling
- Understanding of public cloud resource management such as Alibaba Cloud/AWS
- Familiar with monitoring and alerting systems (Prometheus/Grafana/ELK)

2. Preferred qualifications:
- Experience in operations of large-scale AI training clusters (thousand-GPU level)
- Experience with NCCL/MPI distributed communication libraries and network topology optimization
- Practice with Infrastructure as Code (Terraform/Pulumi)

⚑ 岗位信息收录自公开招聘页（国内官方招聘渠道），以雇主官网为准