AI Infra Engineer(预训练框架与性能优化)
深圳、北京、上海
全职
智能机器人板块
职位描述
面向 VLA 与 World Action Model(视频生成与多模态模型)的预训练需求,参与自研训练框架的开发与演进。通过训练性能分析、分布式执行优化、训练稳定性治理与算法工程协作,提升训练效率与资源利用率,降低训练成本,逐步推动预训练、微调、蒸馏与强化学习共用核心基础能力,支撑不同模型和业务场景的持续扩展。 【职位描述】 1. 训练框架研发:负责预训练框架核心模块的设计、开发与迭代,支持模型、数据、优化器与训练策略的灵活接入;建设统一的训练执行与状态管理机制,完善 checkpoint 保存恢复、指标记录与评测接口,保障训练正确性、可复现性与兼容性; 2. 端到端性能优化:建立性能分析与评测体系,定位数据处理、计算、显存和通信等环节的瓶颈;优化数据加载与预处理流水线、模型计算与显存使用,评估算子优化、编译、低精度等技术的收益,提升 GPU 有效利用率与训练吞吐; 3. 分布式训练与规模扩展:优化并行策略、通信与计算重叠及负载均衡;开展 NCCL 集合通信性能分析与优化,结合 GPU 互联和网络拓扑定位带宽、时延和慢节点问题,完成百卡到千卡规模的扩展验证与容量规划; 4. 训练可靠性与工程效能:建设训练监控、日志、告警与诊断能力,定位并解决 OOM、NaN、卡死等稳定性问题;完善训练状态保存、故障恢复和重试机制,开展故障演练与复盘,降低长任务中断与人工干预; 5. 公共训练能力沉淀:推进分布式执行、资源管理、保存恢复与性能分析等公共能力的抽象与复用,支持预训练与强化学习框架的融合,使视频模型预训练与现有 RL 任务共用核心基础能力。
职位要求
投递

XPeng is a leading Chinese Smart EV company that designs, develops, manufactures, and markets Smart EVs that appeal to the large and growing base of technology-savvy middle-class consumers. Its mission is to drive Smart EV transformation with technology and data, shaping the mobility experience of the future. In order to optimize its customers’ mobility experience, XPeng develops in-house its full-stack advanced driver-assistance system technology and in-car intelligent operating system, as well as core vehicle systems including powertrain and the electrical/electronic architecture. XPeng is headquartered in Guangzhou, China. In 2021, the Company established its European headquarters in Amsterdam, along with other dedicated offices in Copenhagen, Munich, Oslo, and Stockholm.The Company’s Smart EVs are mainly manufactured at its plant in Zhaoqing and Guangzhou,Guangdong province.
For more information, please visit https://heyxpeng.com.