AML-火山方舟大模型推理系统工程师
Core
Design and optimize large-scale training and inference systems for LLMs, managing high-concurrency traffic and heterogeneous hardware clusters.
Role type
Senior IC LLM inference and training systems engineer
Builds
High-performance distributed training clusters and scalable inference systems for LLMs
Domain
Cloud computing / Large Language Models / Distributed Systems
Deliverable
production ML models
Required skills
C/C++, Python, Linux, distributed systems architecture, GPU/NPU/TPU integration, model quantization, subgraph matching, compilation optimization, vLLM, TensorRT-LLM, SGLang, Megatron-LM
Preferred skills
GPU hardware architecture knowledge, CUDA/cuDNN stack expertise, performance analysis, large-scale distributed system design
Responsibilities
Optimize model compute performance and tune thousand-card training clusters; Develop distributed inference systems and manage large-scale traffic scheduling; Research and introduce forward-looking architecture techniques; Integrate heterogeneous hardware into training/inference frameworks; Improve cluster utilization via elastic scheduling and GPU overcommitment; Collaborate with algorithm teams for joint optimization.