Senior GPU Cluster Networking Engineer (RDMA/InfiniBand)

Sciforium; San Francisco (CA); On-site; USD 170,000 - 230,000; Full time

Job summary

Sciforium is building an AI infrastructure platform and needs a Senior Network Engineer to own the networking stack for GPU clusters. You will design, bring up, and operate the complete fabric from InfiniBand/RoCE to data center perimeter and cloud connectivity.

You will lead fabric validation, optimize performance, and manage security, routing, and network automation across multi-vendor environments. This role drives science-grade connectivity for scale.

Qualifications

• 7+ years designing and operating production data center networks, including at least one large-scale HPC/AI cluster fabric (InfiniBand or RoCE v2).
• Expert-level routing and switching: BGP, OSPF, ECMP, VRF across multiple vendor platforms.
• Deep RDMA expertise: RoCE v2 tuning and InfiniBand fabric management.
• Production network security experience: firewalls, VPNs, segmentation.
• Network automation proficiency: Python plus Ansible/Nornir/NAPALM with Git workflows.

Responsibilities

• Fabric Architecture & Cluster Network Design for GPU clusters, including topology and oversubscription analysis.
• RDMA & Performance Engineering: configure RoCE v2 and InfiniBand fabrics, validate with performance tests.
• Production Network Operations & Security: manage routing, perimeter security, NetBox-based config generation, observability.
• Cross-Cluster & Cloud Connectivity: design inter-site links and hybrid cloud connectivity.

Skills

Network design

Routing & switching

RDMA expertise

Security

Network automation

Cloud networking

GPUDirect / NCCL mapping

Tools

Ansible

Nornir

NAPALM

Git / version control

Prometheus/Grafana

Job description

Sciforium is building an AI infrastructure platform and needs a Senior Network Engineer to own the networking stack for GPU clusters. You will design, bring up, and operate the complete fabric from InfiniBand/RoCE to data center perimeter and cloud connectivity.

You will lead fabric validation, optimize performance, and manage security, routing, and network automation across multi-vendor environments. This role drives science-grade connectivity for scale.