阿里云阿里云智能-AI基础设施解决方案架构师-北京/杭州/上海

工作描述

Qualifications
1、计算机科学、电子工程或相关领域本科及以上学历，具有 5 年以上云计算平台、分布式系统、大规模数据处理、AI 基础设施或人工智能系统架构设计与实施经验。
2、深入理解 AI/ML 技术栈，熟悉 PyTorch、TensorFlow 等深度学习框架以及大模型训练和推理的原理与流程；熟悉 DeepSpeed、Megatron-LM 等分布式训练框架及 vLLM、TensorRT-LLM 等推理框架，具备训练或推理性能优化的实际经验。有成功主导大型 AI 项目或产品从 0 到 1 设计与实施经验者优先。
3、精通分布式计算及云原生架构，熟悉 Docker、Kubernetes 等容器化和编排技术；深入了解 GPU、DPU、高性能网络（如 RoCE、InfiniBand）及高性能存储原理，具备大规模 GPU 集群建设、管理或性能调优经验。熟练掌握 Python 等编程语言者优先。
4、熟悉阿里云、AWS、GCP、Azure 等主流云平台，具备云上高可用、高扩展 AI 智算集群的设计和部署经验；能够围绕工作负载特征开展芯片与实例选型、容量规划、网络拓扑和存储方案设计，并在性能、成本和交付效率之间作出合理权衡。熟悉阿里云 PAI、GPU 实例、CPFS 等产品者优先。
5、具备优秀的沟通协调、方案撰写和演讲能力，能够面向客户技术团队、管理层及内部协作团队清晰传达技术决策、架构设计与业务价值；
6、英语须能够作为日常工作语言，并具备良好的跨区域、跨文化协作能力。有海外留学或工作经历，有云产品售前、POC、大型客户项目经验者优先。
7、对 AI 基础设施技术有强烈热情和好奇心，自驱力和学习力强；具备良好的市场敏感度、分析与解决问题能力、团队协作能力和结果意识，善于攻坚复杂问题，能够在快速变化的技术和业务环境中持续推动问题解决与项目落地。

1. Bachelor’s degree or above in Computer Science, Electronic Engineering, or a related field, with 5+ years of experience in cloud platforms, distributed systems, large-scale data processing, AI infrastructure, or the architecture and implementation of AI systems.
2. Deep understanding of the AI/ML technology stack, including deep learning frameworks such as PyTorch and TensorFlow and the principles and workflows of large-model training and inference. Familiarity with distributed training frameworks such as DeepSpeed and Megatron-LM and inference frameworks such as vLLM and TensorRT-LLM, with hands-on experience in training or inference performance optimization. Experience leading the design and implementation of a large-scale AI project or product from zero to one is preferred.
3. Strong expertise in distributed computing and cloud-native architectures, with proficiency in containerization and orchestration technologies such as Docker and Kubernetes. Deep understanding of GPUs, DPUs, high-performance networking such as RoCE and InfiniBand, and high-performance storage. Experience building, managing, or optimizing large-scale GPU clusters is required. Proficiency in programming languages such as Python is preferred.
4. Familiarity with major cloud pla…