Senior Network Architect for 10k+ GPU HPC Clusters

AMD; San Jose (CA); Hybrid; USD 150,000 - 190,000; Full time

Benefits offered by this job

Hybrid work model

AMD benefits

Job summary

AMD is seeking a Senior Network Engineer in San Jose, CA to architect, deploy, and optimize backend networks for expansive GPU clusters. You will own the network from GPU servers to the leaf-spine fabric, ensuring low latency and high bandwidth for AI/HPC workloads.

The role involves design, qualification, and production deployment with cross-team collaboration. You will lead complex initiatives, mentor engineers, and build observability with Prometheus/Grafana, while working in a hybrid

Qualifications

• Bachelor’s degree in Computer Engineering or related field.
• Extensive experience in data center networking for AI/HPC workloads.
• Deep knowledge of RDMA, RoCEv2 and large-scale GPU clusters.

Responsibilities

• Architect and operate high-performance backend networks for large AMD Instinct GPU clusters.
• Design fabrics supporting AI/HPC from rack to 10,000+ GPUs.
• Own backend network from GPU server to leaf-spine fabric.
• Develop RoCEv2-based Ethernet fabrics and high-speed topology.
• Lead incident response and capacity planning for GPU networks.
• Collaborate with AI engineering, data center, storage, and security teams.

Skills

Data center networking

RDMA/RoCEv2

GPU cluster design

Telemetry & performance testing

Leadership & mentoring

Education

Bachelor’s or Master’s in Computer Engineering or related field

Tools

Juniper Junos OS

Prometheus

Grafana

VXLAN/EVPN

Kubernetes

Job description

AMD is seeking a Senior Network Engineer in San Jose, CA to architect, deploy, and optimize backend networks for expansive GPU clusters. You will own the network from GPU servers to the leaf-spine fabric, ensuring low latency and high bandwidth for AI/HPC workloads.

The role involves design, qualification, and production deployment with cross-team collaboration. You will lead complex initiatives, mentor engineers, and build observability with Prometheus/Grafana, while working in a hybrid