Senior GPU Interconnect Reliability Engineer

Firmus; Sydney; On-site; AUD 140,000 - 190,000; Full time

Job summary

Firmus Technologies is seeking a Senior Platform Reliability Engineer, Fabric and Interconnect to own the reliability of GPU interconnect domains, high-performance networks, and DPU-based host networking across the AI FactoryOS estate.

This hands-on role emphasizes automation, with guarded automation and remediation tooling, incident response, vendor escalations, and collaboration with engineering to drive fault closure and scalable uptime.

Qualifications

• 8+ years in production network or interconnect infra.
• Experience with NVLink/NVSwitch and InfiniBand/RoCE.
• Strong automation and IaC experience.
• Ability to act as senior escalation point and runbooks.

Responsibilities

• Ensure reliable operation and automation of GPU interconnect and network fabrics.
• Develop guarded automation and remediation tooling for fabric faults.
• Tune performance across the interconnect stack from application to link level.
• Manage DPU-based host networking including offload paths.
• Execute firmware upgrades and capacity changes with defined rollouts.
• Lead major fabric incident recovery and document runbooks.
• Coordinate with vendors and lead escalations at engineering level.
• Mentor engineers and share best practices across the team.

Skills

High-performance networking

GPU interconnect fabrics (NVLink/NVSw

InfiniBand/RoCE networking

DPU/SmartNIC networking

Automation & IaC

Runbooks/incident response

Vendor escalations & fault diagnosis

Programming/Scripting (Python/Go/Bash)

Tools

Python

Go

Bash

Terraform/IaC tooling

Job description

Firmus Technologies is seeking a Senior Platform Reliability Engineer, Fabric and Interconnect to own the reliability of GPU interconnect domains, high-performance networks, and DPU-based host networking across the AI FactoryOS estate.

This hands-on role emphasizes automation, with guarded automation and remediation tooling, incident response, vendor escalations, and collaboration with engineering to drive fault closure and scalable uptime.