Sr. GPU Cloud East-West Network Expert (SRE SME)
Core
Design, deploy, and operate high-performance InfiniBand and RoCEv2 networks for massive GPU clusters supporting AI cloud and Bitcoin mining workloads.
Role type
Senior Site Reliability Engineer (SRE) specializing in GPU cloud networking
Builds
Scalable GPU clusters (100–10,000 GPUs) and AIOps telemetry pipelines for fault prediction
Domain
Data center networking, High-Performance Computing (HPC), AI Infrastructure
Deliverable
production ML models
Required skills
InfiniBand fabric management, RoCEv2 configuration, NCCL tuning, UFM, fault diagnosis, telemetry-driven operations, runbook-as-code
Preferred skills
400G/800G optics expertise, Nvidia/Mellanox support escalation, adaptive routing, congestion control
Technologies
InfiniBand, RoCEv2, UFM, ibdiagnet, perfquery, ibstat, NCCL, Nvidia, Mellanox, Arista, Cisco
Responsibilities
Manage InfiniBand fabrics (fat-tree, rail-optimized, dragonfly) for GPU clusters; Configure RoCEv2 networks across Nvidia, Arista, and Cisco platforms; Monitor and tune IB/RoCE performance including adaptive routing and congestion control; Feed telemetry into AIOps substrate for fault prediction; Diagnose network faults like link flaps and packet drops; Coordinate with vendor support for escalations and RMAs.
Seniority
Senior, hands-on IC