GPU异构服务器架构师
Core
Design and optimize heterogeneous GPU hardware systems and architectures for AI and rendering workloads, serving Tencent's internal business applications.
Role type
Senior IC GPU heterogeneous server architect
Builds
Heterogeneous GPU hardware products and system architectures
Domain
Hardware engineering, AI infrastructure, High-performance computing
Deliverable
production ML models | infrastructure
Required skills
Heterogeneous computing architecture, GPU hardware system design, AI training framework optimization, Parallel computing network architecture, Performance benchmarking, Cross-functional project leadership
Preferred skills
FPGA integration, PCIe/DDR/storage protocol expertise, Cost-performance optimization
Technologies
TensorFlow, PyTorch, CUDA, OpenCL, PCIe, DDR
Responsibilities
Define hardware system requirements and product roadmaps for heterogeneous GPU applications; Design and optimize overall hardware architectures; Integrate cross-domain teams to meet product specifications; Evaluate GPU performance metrics (bandwidth, memory, compute, power) for AI and rendering; Optimize training frameworks and test performance across different GPU vendors; Modify and tune collective communication libraries for parallel computing; Collaborate with OS and business teams to innovate hardware solutions; Analyze component failures to guide future architecture design.
Seniority
Senior, hands-on IC
