GPU Networking Architect for Datacenter & HPC

Prime Intellect, Inc.; San Francisco (CA); On-site; USD 150,000 - 300,000; Full time

Job summary

Prime Intellect, Inc. is seeking an experienced network engineer to design and operate the networks that connect large GPU clusters.

You will own the reliability and performance of training fabrics, storage networks, and management connectivity, enabling distributed workloads to scale without bottlenecks. You will collaborate with datacenter operators, hardware vendors, and ML teams to ensure high availability, performance, and smooth deployments across frontier AI workloads.

Qualifications

• 3+ years of production datacenter networking experience.
• Strong Ethernet, TCP/IP, routing, switching knowledge.
• Hands-on GPU networking with InfiniBand or RoCE.

Responsibilities

• Design and deploy datacenter network topologies for GPU training/inference.
• Configure Ethernet/RoCE and InfiniBand fabrics with routing standards.
• Automate provisioning, validation, upgrades, and rollbacks.
• Diagnose packet loss, congestion, link failures, and performance issues.
• Benchmark network performance with infra and ML teams.
• Build port health monitoring and incident response playbooks.

Skills

Datacenter networking

Ethernet/TCP/IP

GPU networking (InfiniBand/RoCE)

Linux troubleshooting

Python/automation

Network monitoring

Education

Bachelor's in CS/EE or related field

Tools

Python

Ansible

Linux

NVIDIA/NIC hardware

Job description

Prime Intellect, Inc. is seeking an experienced network engineer to design and operate the networks that connect large GPU clusters.

You will own the reliability and performance of training fabrics, storage networks, and management connectivity, enabling distributed workloads to scale without bottlenecks. You will collaborate with datacenter operators, hardware vendors, and ML teams to ensure high availability, performance, and smooth deployments across frontier AI workloads.