Senior Network Engineer (InfiniBand & HPC Architecture)

Axe Compute is seeking a Senior Network Engineer to own, optimize, and troubleshoot the InfiniBand and HPC interconnect fabrics powering its GPU infrastructure. The role focuses on fabric design and operations, complex network escalations, storage transport optimization, telemetry integration, operational runbooks, and vendor coordination.

Responsibilities:

• InfiniBand Fabric Governance & Optimization: Own operational health, topology design, Subnet Manager (OpenSM / UFM) tuning, Adaptive Routing, Congestion Control (CC), and firmware lifecycle for NDR/HDR Mellanox InfiniBand fabrics
• Tier-3/4 Complex Escalations: Serve as the ultimate technical escalation point for fabric bottlenecks, link flap events, BER (Bit Error Rate) optical degradations, and latency spikes across active GPU training clusters
• Storage Transport Optimization: Fine-tune RoCEv2 and InfiniBand transport parameters to guarantee sustained multi-terabit throughput between compute nodes, switch fabrics, and distributed storage systems
• Telemetry & Automation Integration: Partner with Sr. Automation Engineers to stream gNMI, SNMP, and InfiniBand performance counters into Grafana dashboards, PagerDuty, and ServiceNow for automated alerting and root-cause suppression
• Operational SOPs & Field Runbooks: Author step-by-step diagnostic and hot-swap runbooks for optical cables, transceivers, and switch modules for 24/7 global NOC teams (US & APAC) to execute during shift coverage
• Vendor & OEM Escalation: Lead technical engagement with NVIDIA/Mellanox TAC, hardware OEMs, optical vendors, and data center smart hands during major incident root-cause analysis (RCA)

Requirements:

• 6+ years in enterprise data center networking, HPC interconnects, or ISP backbones, with 3+ years dedicated to high-speed InfiniBand fabric operations
• Deep expert command of NVIDIA/Mellanox InfiniBand architecture, Subnet Managers (UFM/OpenSM), ibdiagnet, ibstat, perfquery, and fabric topology validation
• Strong proficiency with BGP, EVPN-VXLAN, Spine-Leaf architectures, RoCEv2, L2/L3 switching, and optical transceiver diagnostics across Arista, Cisco, or Juniper platforms
• Advanced Linux CLI fluency (tcpdump, ethtool, ipmitool, dmesg, iperf3, lspci) for host-level HCA/NIC and driver debugging
• Proven track record driving rapid MTTR during critical (P1/P2) network events on high-density compute infrastructure
• Direct experience supporting multi-cluster GPU/AI training environments (NVIDIA H100/H200/B200 platforms)
• NVIDIA Certified Professional/Expert (InfiniBand or Ethernet), Arista ACE, or CCNP/CCIE
• Network automation experience using Python, Ansible, or Terraform for fabric configuration management