机器学习平台调度工程师（北京/深圳）
Tencent
机器学习平台调度工程师（北京/深圳）
Job Description
机器学习平台调度工程师（北京/深圳）
📋 Job Overview
The Machine Learning Platform Scheduling Engineer role at Tencent involves leading the global resource scheduling for large-scale GPU clusters to optimize resource utilization and ensure efficient operation of offline and online tasks. The position focuses on enhancing performance through optimizations in RDMA networks, distributed storage, and compute resources, while building high-availability scheduling frameworks using Kubernetes and Docker. Responsibilities also include exploring advanced technologies like hybrid cloud, virtualization, and ARM heterogeneous computing to drive platform innovation.
📍 Location: Beijing, China
🏢 Business Unit: TEG
📄 Full Description
2.深入优化RDMA高速网络、分布式存储与计算资源的协同调度，有效解决大规模训练任务中的性能瓶颈，提升整体计算效率；
3.基于Kubernetes、Docker等云原生技术，构建高可用调度框架，全面支持分布式训练框架，实现任务编排、容灾与混部能力，并深入K8s调度器、CSI插件及CRD的开发，推动大规模训推技术的实际落地；
4.积极探索混合云、虚拟化、ARM异构计算等前沿方向，不断推动技术与平台能力的升级和创新。
🎯 Key Responsibilities
- Lead global resource scheduling for 10,000-card GPU clusters through refined management and optimization strategies to significantly improve resource utilization and ensure efficient, stable operation of offline and online tasks.
- Deeply optimize the collaborative scheduling of RDMA high-speed networks, distributed storage, and compute resources to effectively resolve performance bottlenecks in large-scale training tasks and enhance overall computational efficiency.
- Build high-availability scheduling frameworks based on Kubernetes, Docker, and other cloud-native technologies to fully support distributed training frameworks, enabling task orchestration, disaster recovery, and mixed deployment capabilities, while developing K8s schedulers, CSI plugins, and CRDs to promote the practical implementation of large-scale training and inference technologies.
- Actively explore cutting-edge directions such as hybrid cloud, virtualization, and ARM heterogeneous computing to continuously drive upgrades and innovations in technology and platform capabilities.
🛠️ Required Skills
- Expertise in GPU cluster resource scheduling and optimization
- Knowledge of RDMA high-speed networks and distributed storage
- Proficiency in Kubernetes, Docker, and cloud-native technologies
- Experience with distributed training frameworks and K8s components (schedulers, CSI plugins, CRDs)
- Familiarity with hybrid cloud, virtualization, and ARM heterogeneous computing
Locations
- Beijing, China
Salary
300,000 - 600,000 CNY / yearly
Source: ai estimated
* This is an estimated range based on market data and may vary based on experience and qualifications.
Skills Required
- Expertise in GPU cluster resource scheduling and optimizationintermediate
- Knowledge of RDMA high-speed networks and distributed storageintermediate
- Proficiency in Kubernetes, Docker, and cloud-native technologiesintermediate
- Experience with distributed training frameworks and K8s components (schedulers, CSI plugins, CRDs)intermediate
- Familiarity with hybrid cloud, virtualization, and ARM heterogeneous computingintermediate
Responsibilities
- Lead global resource scheduling for 10,000-card GPU clusters through refined management and optimization strategies to significantly improve resource utilization and ensure efficient, stable operation of offline and online tasks.
- Deeply optimize the collaborative scheduling of RDMA high-speed networks, distributed storage, and compute resources to effectively resolve performance bottlenecks in large-scale training tasks and enhance overall computational efficiency.
- Build high-availability scheduling frameworks based on Kubernetes, Docker, and other cloud-native technologies to fully support distributed training frameworks, enabling task orchestration, disaster recovery, and mixed deployment capabilities, while developing K8s schedulers, CSI plugins, and CRDs to promote the practical implementation of large-scale training and inference technologies.
- Actively explore cutting-edge directions such as hybrid cloud, virtualization, and ARM heterogeneous computing to continuously drive upgrades and innovations in technology and platform capabilities.
Tags & Categories
Answer 10 quick questions to check your fit for 机器学习平台调度工程师（北京/深圳） @ Tencent.
Related Books and Jobs
机器学习平台调度工程师（北京/深圳）
Tencent
机器学习平台调度工程师（北京/深圳）
Job Description
机器学习平台调度工程师（北京/深圳）
📋 Job Overview
The Machine Learning Platform Scheduling Engineer role at Tencent involves leading the global resource scheduling for large-scale GPU clusters to optimize resource utilization and ensure efficient operation of offline and online tasks. The position focuses on enhancing performance through optimizations in RDMA networks, distributed storage, and compute resources, while building high-availability scheduling frameworks using Kubernetes and Docker. Responsibilities also include exploring advanced technologies like hybrid cloud, virtualization, and ARM heterogeneous computing to drive platform innovation.
📍 Location: Beijing, China
🏢 Business Unit: TEG
📄 Full Description
2.深入优化RDMA高速网络、分布式存储与计算资源的协同调度，有效解决大规模训练任务中的性能瓶颈，提升整体计算效率；
3.基于Kubernetes、Docker等云原生技术，构建高可用调度框架，全面支持分布式训练框架，实现任务编排、容灾与混部能力，并深入K8s调度器、CSI插件及CRD的开发，推动大规模训推技术的实际落地；
4.积极探索混合云、虚拟化、ARM异构计算等前沿方向，不断推动技术与平台能力的升级和创新。
🎯 Key Responsibilities
- Lead global resource scheduling for 10,000-card GPU clusters through refined management and optimization strategies to significantly improve resource utilization and ensure efficient, stable operation of offline and online tasks.
- Deeply optimize the collaborative scheduling of RDMA high-speed networks, distributed storage, and compute resources to effectively resolve performance bottlenecks in large-scale training tasks and enhance overall computational efficiency.
- Build high-availability scheduling frameworks based on Kubernetes, Docker, and other cloud-native technologies to fully support distributed training frameworks, enabling task orchestration, disaster recovery, and mixed deployment capabilities, while developing K8s schedulers, CSI plugins, and CRDs to promote the practical implementation of large-scale training and inference technologies.
- Actively explore cutting-edge directions such as hybrid cloud, virtualization, and ARM heterogeneous computing to continuously drive upgrades and innovations in technology and platform capabilities.
🛠️ Required Skills
- Expertise in GPU cluster resource scheduling and optimization
- Knowledge of RDMA high-speed networks and distributed storage
- Proficiency in Kubernetes, Docker, and cloud-native technologies
- Experience with distributed training frameworks and K8s components (schedulers, CSI plugins, CRDs)
- Familiarity with hybrid cloud, virtualization, and ARM heterogeneous computing
Locations
- Beijing, China
Salary
300,000 - 600,000 CNY / yearly
Source: ai estimated
* This is an estimated range based on market data and may vary based on experience and qualifications.
Skills Required
- Expertise in GPU cluster resource scheduling and optimizationintermediate
- Knowledge of RDMA high-speed networks and distributed storageintermediate
- Proficiency in Kubernetes, Docker, and cloud-native technologiesintermediate
- Experience with distributed training frameworks and K8s components (schedulers, CSI plugins, CRDs)intermediate
- Familiarity with hybrid cloud, virtualization, and ARM heterogeneous computingintermediate
Responsibilities
- Lead global resource scheduling for 10,000-card GPU clusters through refined management and optimization strategies to significantly improve resource utilization and ensure efficient, stable operation of offline and online tasks.
- Deeply optimize the collaborative scheduling of RDMA high-speed networks, distributed storage, and compute resources to effectively resolve performance bottlenecks in large-scale training tasks and enhance overall computational efficiency.
- Build high-availability scheduling frameworks based on Kubernetes, Docker, and other cloud-native technologies to fully support distributed training frameworks, enabling task orchestration, disaster recovery, and mixed deployment capabilities, while developing K8s schedulers, CSI plugins, and CRDs to promote the practical implementation of large-scale training and inference technologies.
- Actively explore cutting-edge directions such as hybrid cloud, virtualization, and ARM heterogeneous computing to continuously drive upgrades and innovations in technology and platform capabilities.
Tags & Categories
Answer 10 quick questions to check your fit for 机器学习平台调度工程师（北京/深圳） @ Tencent.
Related Books and Jobs