What you will be doing:
Tier 3 support for InfiniBand and Ethernet fabrics: installing, supporting, and resolving complex technical issues in large AI/HPC and storage clusters.
Troubleshooting end-to-end InfiniBand fabrics – subnet management (UFM / OpenSM), routing topology and congestion control.
Developing and presenting comprehensive technical solutions to customer problems, including project management of complex fabric installations.
Problem reporting, issue replication, and resolution management.
Responding to customer technical inquiries.
Site visits and conference calls with customers and partners.
Developing and refining internal processes to improve support efficiency and productivity.
What we need to see:
B.Sc. in Computer Science, Software Engineering, or Electrical Engineering, or equivalent experience.
5+ years providing customer support for hardware & software products.
Hands-on InfiniBand experience – OR deep Ethernet/RDMA expertise with a demonstrated ability and strong motivation to ramp quickly on InfiniBand (subnet manager, fabric topologies such as fat-tree, RDMA verbs, diagnostics).
Extensive knowledge of LAN switching/routing (STP, MSTP, MLAG, VPC, VRRP, LACP) and IP routing (OSPF, BGP, PIM), plus virtualization, EVPN, VXLAN.
Linux experience and scripting.
Automation – Ansible.
Ability to work under pressure and support high-level customers.
Experience operating and configuring major vendors' switches and routers.
Analysis and diagnosis of highly complex networking problems.
Strong communication, presentation, and oral skills; excellent verbal and written English.
Ways to stand out from the crowd:
Deep InfiniBand expertise: UFM, OpenSM, adaptive routing, congestion control, link-layer troubleshooting, ibdiagnet/perftest, ConnectX/BlueField adapters.
RDMA, RoCE, and GPUDirect RDMA in production AI/HPC clusters.
Clustering and data-center technologies, including upper-layer protocols (e.g., MPI, NCCL).
Knowledge of Linux/Unix at an administration level.
Ethernet (10/40/100+ GigE) and/or InfiniBand at scale.
