高性能网络通信库研发工程师
Tencent
高性能网络通信库研发工程师
Job Description
高性能网络通信库研发工程师
📋 Job Overview
Tencent is seeking a High-Performance Network Communication Library R&D Engineer to develop and optimize communication operators for large model training and inference scenarios. The role involves designing efficient collective communication primitives, implementing computation-communication overlap strategies, and tuning performance across hardware platforms like GPU and NPU. Additionally, the engineer will contribute to the overall architecture design of the AI communication library, establishing technical solutions and development standards.
📍 Location: Shenzhen, China
🏢 Business Unit: TEG
📄 Full Description
2.通信计算融合：设计通信与计算的overlap策略，实现计算通信并行，提升大模型整体训练/推理性能；
3.通信性能调优：针对不同硬件平台（GPU、NPU等）进行算子性能分析和优化，解决性能瓶颈问题；
4.参与AI通信库整体架构设计，制定技术方案和开发规范。
🎯 Key Responsibilities
- Design and implement efficient collective communication operators (AllReduce, AllGather, ReduceScatter, AllToAll, etc.) for large model training/inference scenarios.
- Design communication-computation overlap strategies to enable parallel computation and communication, enhancing overall large model training/inference performance.
- Perform operator performance analysis and optimization for different hardware platforms (GPU, NPU, etc.), resolving performance bottlenecks.
- Participate in the overall architecture design of the AI communication library, formulating technical solutions and development standards.
Locations
- Shenzhen, China
Salary
300,000 - 800,000 CNY / yearly
Source: ai estimated
* This is an estimated range based on market data and may vary based on experience and qualifications.
Responsibilities
- Design and implement efficient collective communication operators (AllReduce, AllGather, ReduceScatter, AllToAll, etc.) for large model training/inference scenarios.
- Design communication-computation overlap strategies to enable parallel computation and communication, enhancing overall large model training/inference performance.
- Perform operator performance analysis and optimization for different hardware platforms (GPU, NPU, etc.), resolving performance bottlenecks.
- Participate in the overall architecture design of the AI communication library, formulating technical solutions and development standards.
Tags & Categories
高性能网络通信库研发工程师
Tencent
高性能网络通信库研发工程师
Job Description
高性能网络通信库研发工程师
📋 Job Overview
Tencent is seeking a High-Performance Network Communication Library R&D Engineer to develop and optimize communication operators for large model training and inference scenarios. The role involves designing efficient collective communication primitives, implementing computation-communication overlap strategies, and tuning performance across hardware platforms like GPU and NPU. Additionally, the engineer will contribute to the overall architecture design of the AI communication library, establishing technical solutions and development standards.
📍 Location: Shenzhen, China
🏢 Business Unit: TEG
📄 Full Description
2.通信计算融合：设计通信与计算的overlap策略，实现计算通信并行，提升大模型整体训练/推理性能；
3.通信性能调优：针对不同硬件平台（GPU、NPU等）进行算子性能分析和优化，解决性能瓶颈问题；
4.参与AI通信库整体架构设计，制定技术方案和开发规范。
🎯 Key Responsibilities
- Design and implement efficient collective communication operators (AllReduce, AllGather, ReduceScatter, AllToAll, etc.) for large model training/inference scenarios.
- Design communication-computation overlap strategies to enable parallel computation and communication, enhancing overall large model training/inference performance.
- Perform operator performance analysis and optimization for different hardware platforms (GPU, NPU, etc.), resolving performance bottlenecks.
- Participate in the overall architecture design of the AI communication library, formulating technical solutions and development standards.
Locations
- Shenzhen, China
Salary
300,000 - 800,000 CNY / yearly
Source: ai estimated
* This is an estimated range based on market data and may vary based on experience and qualifications.
Responsibilities
- Design and implement efficient collective communication operators (AllReduce, AllGather, ReduceScatter, AllToAll, etc.) for large model training/inference scenarios.
- Design communication-computation overlap strategies to enable parallel computation and communication, enhancing overall large model training/inference performance.
- Perform operator performance analysis and optimization for different hardware platforms (GPU, NPU, etc.), resolving performance bottlenecks.
- Participate in the overall architecture design of the AI communication library, formulating technical solutions and development standards.
Tags & Categories