> Markdown version of [/jobs/ext/4586629-network-engineer](https://www.wearedevelopers.com/jobs/ext/4586629-network-engineer). Every page supports `.md` or `Accept: text/markdown`. Links point to the HTML versions so they work for humans too. Agent guide: [/agents.md](https://www.wearedevelopers.com/agents.md).

---

# Network Engineer

- **Company:** Andromeda Cluster, Inc.
- **Location:** San Francisco, CA, United States (Remote available)
- **Contract:** Permanent contract
- **Skills:** Artificial Intelligence, Computer Clusters, Data Centers, Ethernet, InfiniBand, Storage Area Network (SAN), Python (Programming Language), Network Planning and Design, Ansible, Graphics Processing Unit (GPU), Performance Testing, Machine Learning Operations, Front End Software Development
- **Published:** November 5, 2026
- **Apply:** https://startup.jobs/network-engineer-andromeda-cluster-10682792

## About the Role

* GPU fabrics at scale: Hands-on design and bring-up of InfiniBand or RoCE fabrics for clusters of 1,000+ GPUs.
 * Tooling: Experience with NVIDIA UFM and switch operating systems such as Cumulus or SONiC.
 * Physical layer: Practical knowledge of optics, transceivers, and high-density cabling.
 * Automation: Python and Ansible or similar for configuration and validation.

Strong Candidates May Have

 * NVL72: Experience with GB200 or GB300 NVL72 rack-scale systems.
 * Multi-site: Data center interconnect and cloud interconnect design.
 * Customer-facing: Experience working directly with customer ML infrastructure teams.

## Description

Andromeda's infrastructure team designs, deploys, and operates large-scale GPU clusters for AI labs and compute providers. The Network Engineer designs and brings up the networks in our GPU clusters: compute fabric, storage network, front-end, and out-of-band management. You work with the Cluster Architect on design and with deployment engineers on site through validation. You report to the Head of Infrastructure., * Compute fabric: InfiniBand (NDR, XDR) and Spectrum-X Ethernet fabrics, including rail-optimized topologies and NVLink domain integration.

 * Storage and front-end networks: Design and capacity planning for storage traffic, tenant access, and internet and cloud interconnect.
 * Cabling and optics: Cabling plans, optics BOMs, and labeling standards integrators can build from.
 * Bring-up and validation: Fabric bring-up, NCCL and fabric performance testing, and sign-off for acceptance.
 * Troubleshooting: Escalation point for fabric issues during deployment and in Day 2.
 * Automation: Configuration templates and tooling so each build starts from a known-good baseline.