Senior InfiniBand HPC Networking Architect – Remote

Energy Jobline ZR; Miami (FL); Hybrid; USD 185,000 - 200,000; Full time

Job summary

Axe Compute is seeking a senior InfiniBand & HPC Network Engineer to own, optimize, and troubleshoot the ultra-low-latency interconnect fabrics powering our GPU supercomputing infrastructure.

You will be the SME for high-speed fabric design, Subnet Manager tuning, optical diagnostics, and Tier-3/4 escalations across multi-site data centers, collaborating with the Network Support Manager, Storage Leads, and global NOC teams to maximize throughput and zero packet drop.

Qualifications

• 6+ years in enterprise data center networking or HPC interconnects.
• 3+ years in high-speed InfiniBand fabric operations.
• Experience with Subnet Managers (OpenSM/UFM) and OpenFabrics.
• Strong troubleshooting of latency, BER, and throughput.

Responsibilities

• Own operational health and topology design for InfiniBand fabrics.
• Tune Subnet Manager, Adaptive Routing, and CC; manage firmware lifecycle.
• Escalate and resolve tier-3/4 fabric issues across multi-site data centers.
• Optimize RoCEv2 and InfiniBand transport for multi-terabit throughput.
• Integrate telemetry with Grafana, PagerDuty, ServiceNow for alerts.
• Document SOPs and runbooks for 24/7 NOC coverage.

Skills

InfiniBand mastery

Linux CLI

Incident leadership

BGP & EVPN-VXLAN

Network automation

Education

NVIDIA Certified Professional/Expert

Arista ACE

CCNP/CCIE

Tools

Python

Ansible

Terraform

ibdiagnet

ibstat

tcpdump

iperf3

ethtool

Job description

Axe Compute is seeking a senior InfiniBand & HPC Network Engineer to own, optimize, and troubleshoot the ultra-low-latency interconnect fabrics powering our GPU supercomputing infrastructure.

You will be the SME for high-speed fabric design, Subnet Manager tuning, optical diagnostics, and Tier-3/4 escalations across multi-site data centers, collaborating with the Network Support Manager, Storage Leads, and global NOC teams to maximize throughput and zero packet drop.