Network SRE Expert

Job Summary

To support the development of an AI-operated GPU cloud, the full-time Network SRE Expert will manage InfiniBand and RoCEv2 networks, ensuring optimal performance and reliability for GPU clusters while working remotely from San Jose, CA or Austin, TX.

Key responsibilities

• Oversee the deployment and management of InfiniBand fabrics and RoCEv2 networks for GPU clusters, ensuring efficient data flow and performance tuning
• Implement monitoring and diagnostics using UFM and other tools to identify and resolve network issues, including link degradation and congestion
• Collaborate with platform teams to enhance AIOps capabilities by integrating telemetry data and developing predictive maintenance workflows

Required qualifications

• 5+ years of experience in data center networking, with a minimum of 3 years focused on InfiniBand or RoCE fabrics
• Hands-on experience with Nvidia/Mellanox InfiniBand switches and RoCEv2 deployment, including relevant configurations
• Strong understanding of IB subnet management, QoS, and performance monitoring tools
• Proficiency in diagnosing network issues using tools such as ibdiagnet and perfquery
• Familiarity with NCCL and its implications for GPU communication and network topology