AI系统性能工程师-国际化内容安全平台
Core
Design and develop a next-generation multimodal inference architecture and high-performance computing infrastructure for a massive online traffic content safety platform.
Role type
Senior IC AI System Performance Engineer (Multimodal Inference)
Builds
High-throughput, low-latency distributed inference systems for 200B+ parameter models and VLMs.
Domain
AI Infrastructure / Large Language Models / Multimodal Systems
Deliverable
production ML models
Required skills
Distributed system architecture, High-concurrency scheduling, GPU microarchitecture optimization, CUDA/Triton kernel development, vLLM/SGLang framework internals, Multimodal model engineering (VLM), Quantization (PTQ/QAT), KV Cache optimization.
Preferred skills
AI Agent for system tuning, MoE architecture optimization, New Attention mechanisms, Triton/CUTLASS usage.
Responsibilities
Design global scheduling systems for heterogeneous compute resources; Optimize distributed inference strategies (TP/EP/DP); Develop high-performance low-level operators; Build AI-driven infrastructure for deployment and consistency checks.
Seniority
Senior, hands-on IC
