星星海实验室-AI Infra 研发工程师-GPU方向
Core
Evaluate and optimize performance of AI acceleration hardware across single-node/single-GPU and super-node scenarios, ensuring software-hardware system tuning and business delivery.
Role type
AI infrastructure R&D engineer (GPU direction)
Builds
Optimized AI training and inference systems for hardware products
Domain
AI infrastructure, GPU computing, high-performance computing
Deliverable
production ML models
Required skills
Linux development, Python, C++, GPU architecture knowledge, distributed training frameworks (Megatron), high-performance network communication, inference frameworks (TensorRT, vLLM, SGLang)
Preferred skills
System design optimization, problem-solving, teamwork
Technologies
Linux, Python, C++, Nvidia GPUs, AMD GPUs, Megatron, TensorRT, vLLM, SGLang, high-performance networks
Responsibilities
Conduct comprehensive software-hardware performance evaluation and tuning for AI acceleration hardware; continuously optimize product line quality and user experience; track, locate, and resolve technical issues in collaboration with the team.
