高性能通信互联研发工程师（J102926）
Core
Optimize Scale-up and Scale-out interconnects for high-performance computing nodes, focusing on RDMA-based communication and large model inference performance.
Role type
Senior IC high-performance computing interconnect engineer
Builds
High-performance computing clusters and large language model inference systems
Domain
High-performance computing + Distributed systems
Deliverable
production ML models
Required skills
RDMA protocol (RoCEv2/InfiniBand), C/C++, Python, RNIC tuning, collective communication optimization, distributed parallel strategies (TP/PP/EP), KV Cache management
Preferred skills
SGLang, vLLM, IBGDA, DeepEP, KunlunXin hardware deployment
Technologies
RoCEv2, InfiniBand, RNIC, SGLang, vLLM, DeepEP, IBGDA, KunlunXin
Responsibilities
Optimize interconnect performance for 32-256 card Scale-up and cross-node Scale-out RDMA; Tune RNIC congestion control and parameters; Analyze bandwidth/latency bottlenecks in All-Reduce/All-to-All operations; Deploy and validate large models on inference engines; Identify link anomalies and drive optimization solutions
Seniority
Intern, 6 months
