异构适配与优化AI Agent研发工程师/专家（深圳/北京/上海）
Core
Develop AI agents to automate operator development, cross-chip migration, and deep optimization for large model inference frameworks.
Role type
Senior IC AI Infrastructure Engineer (LLM Agents & High-Performance Computing)
Builds
Automated operator generation agents, cross-chip migration agents, and optimized inference frameworks (vLLM/SGLang) for specific hardware.
Domain
AI Infrastructure / High-Performance Computing / Large Language Models
Deliverable
production ML models
Required skills
LLM Agent development (Prompt Engineering, Tool Use, Code Generation), CUDA or Triton, PyTorch operator mechanisms, vLLM/SGLang frameworks, Profiling and performance tuning, CI/CD and automated testing, Cross-chip adaptation
Preferred skills
AI Infra / High-Performance Computing experience, AI operator generation projects, Experience with specific chip adaptation
Technologies
vLLM, SGLang, CUDA, Triton, PyTorch, LLMs
Responsibilities
Build operator development agents for code generation and verification; Build cross-chip migration agents for GPU-to-chip adaptation; Adapt frameworks like vLLM/SGLang to target chips and optimize key paths (MoE, Attention, KV Cache); Perform performance analysis and tuning across compute, memory, and communication dimensions; Codify expert knowledge into agent tools for automated optimization.
