AI Infra工程师（芯片平台） - 芯片研发
Core
Building a full-stack AI infrastructure base covering AI compute clusters, network storage, heterogeneous compute scheduling, security operations, cost, and performance governance, plus an enterprise-grade LLMOps engineering system.
Role type
Senior AI Infrastructure Engineer (LLMOps & Compute Platform)
Builds
AI compute clusters, LLMOps pipelines, model deployment services, and high-performance networking for large models.
Domain
AI Infrastructure / Large Language Model Operations / Chip Platform
Deliverable
production ML models
Required skills
GPU cluster construction and operations, model deployment and tuning, API router scheduling, distributed KV Cache storage and offloading, high-speed network architecture (TCP/IP, IB, RoCE v2), model performance tuning, token resource quota management, model performance evaluation.
Preferred skills
Experience with enterprise large model infrastructure, AI compute cluster implementation, self-developed or secondarily developed model gateways.
Technologies
vLLM, SGLang, TCP/IP, IB, RoCE v2
Responsibilities
Design and build full-stack AI infrastructure, establish LLMOps engineering systems, manage GPU cluster load balancing, optimize network performance, and handle model lifecycle management.
Seniority
Senior, hands-on IC
