Senior GPU Fabric Network Engineer (Remote, 25% Travel)

Andromeda Cluster; San Francisco (CA); Hybride; USD 120 000 - 180 000; Plein temps

Résumé du poste

Andromeda Cluster in San Francisco is seeking a Network Engineer to design, deploy, and operate multi-site GPU clusters for AI workloads. You will own compute fabric, storage networks, front-end, and out-of-band management, collaborating with the Cluster Architect and deployment engineers on site.

The role emphasizes InfiniBand/RoCE fabrics, NVIDIA UFM tooling, and automation using Python and Ansible to start from a known-good baseline. Travel up to 25% and a remote-friendly US schedule.

Qualifications

• Experience designing and bringing up large GPU fabrics.
• Familiarity with InfiniBand or RoCE in 1,000+ GPU clusters.
• Strong automation with Python and Ansible.
• Ability to work with on-site deployment teams.

Responsabilités

• Design and bring up compute fabric for GPU clusters.
• Plan storage and front-end networks for performance and scale.
• Develop cabling plans, optics BOMs, and labeling standards.
• Fabric bring-up, NCCL and fabric performance testing, sign-off for acceptance.
• Troubleshoot fabric issues during deployment and Day 2 operations.
• Create configuration templates and tooling to start builds from a known-good baseline.

Connaissances

Python

InfiniBand

RoCE fabrics

Automation

Outils

NVIDIA UFM

Cumulus/SONiC

Switch OS

Description du poste

Andromeda Cluster in San Francisco is seeking a Network Engineer to design, deploy, and operate multi-site GPU clusters for AI workloads. You will own compute fabric, storage networks, front-end, and out-of-band management, collaborating with the Cluster Architect and deployment engineers on site.

The role emphasizes InfiniBand/RoCE fabrics, NVIDIA UFM tooling, and automation using Python and Ansible to start from a known-good baseline. Travel up to 25% and a remote-friendly US schedule.