# AI平台研发工程师

External opening: you cannot apply through AI Native Match; apply on the official site.

Apply through the official channel. The full external receipt stays in your workspace. When signed in, three minimal event types explicitly recorded through the CLI are reported by default: selection, self-reported application and channel observation. Browsing or saving a role does not report an event; external applications do not appear under My applications.

- Company: 零一万物 (Startup)
- Locations: 北京
- Role: Engineering
- Apply on the official site: https://01ai.jobs.feishu.cn/index/position/7685963157332330798/detail

## Role description from source

1、 基于 Kubernetes 建设多租户 GPU 算力调度平台，支持弹性伸缩与任务级管理
2、实现 GPU 拓扑感知调度（NVLink/PCIe），降低跨卡通信 P99 尾延迟
3、 探索 MIG、时间分片等 GPU 切分方案，提升卡级利用率
4、构建高并发大模型在线推理平台
5、 建设 GPU、网络、存储、推理全链路可观测性体系
6、 推进 Spot 实例与自动扩缩容，降低算力成本
7、 建立故障定位、应急响应与根因分析机制
职位要求
1、熟练 Golang、Python
2、熟悉 Kubernetes、Docker，有调度框架二次开发经验优先
3、熟悉至少一种训练/推理框架：PyTorch、Megatron、DeepSpeed、vLLM、SGLang
4、理解 GPU 架构、CUDA、异构计算；了解 RDMA、NVLink、InfiniBand
加分项：K8s Scheduler Framework、Volcano/Kueue、PD 分离、KV Cache、AI Agent 基础设施等

---
opening_id: 7dcd0bfc-818e-4626-ade6-50a5669dc2b8

`anm openings save --id 7dcd0bfc-818e-4626-ade6-50a5669dc2b8`