大模型可观测/模型优化工程师（J105368）
Core
Build an observability platform for large language models covering the full lifecycle of training and inference to analyze performance, accuracy, and stability.
Role type
Senior IC AI Infrastructure Engineer (LLM Observability & Optimization)
Builds
AI Infra observability platform, model analysis tools, and optimization toolchains
Domain
AI Infrastructure, Large Language Models, GPU/Heterogeneous Computing
Deliverable
production ML models
Required skills
Python, C++, Linux, PyTorch, CUDA, GPU programming, model quantization, model adaptation, performance profiling, observability platform development
Preferred skills
NPU/XPU experience, training/inference framework knowledge, fault localization
Technologies
PyTorch, CUDA, Linux, GPU, NPU, XPU
Responsibilities
Develop tools for model adaptation, performance optimization, and accuracy issue localization; build metrics, log, trace, and profiling capabilities for GPU/heterogeneous compute environments.
Seniority
Senior, hands-on IC
