Staff HPC Network Architect

Lambda; United States; Hybrid; USD 180,000 - 280,000; Full time

Job summary

Lambda, The Superintelligence Cloud, seeks a senior network architect to build and scale ultra-low-latency data-center networks for AI workloads. You will design GPU-cluster topologies, multi-tenant fabrics, and high-bandwidth paths, collaborating with compute and storage teams to ensure fault-tolerant data flow across sites.

You have 7+ years in this field, deep InfiniBand and Ethernet fabric expertise, and strong leadership.

Qualifications

• 7+ years architecting high-performance data center networks.
• Deep expertise with InfiniBand HDR/NDR and Ethernet fabrics (RoCE/RDMA).
• Strong understanding of data center switching architectures, QoS, and network virtualization.

Responsibilities

• Architect high-performance networking solutions for ultra-low latency and high bandwidth.
• Define topology and architectural patterns for large-scale GPU clusters and multi-tenant environments.
• Evaluate, benchmark, and select next-generation network technologies to meet AI workload requirements.
• Develop and maintain network architecture standards, reference designs, and scalability roadmaps for multi-site and hybrid environments.
• Partner with compute and storage architects to ensure seamless end-to-end data flow and fault tolerance.
• Guide network automation strategies and tooling for provisioning, telemetry, and observability.
• Mentor engineers on advanced network concepts, troubleshooting, and best practices.

Skills

High-performance networking

HPC/AI infra

Latency optimization

BGP EVPN

RDMA and InfiniBand

Automation & telemetry

Leadership

Tools

InfiniBand HDR/NDR

RoCE

RDMA

VXLAN

EVPN

Job description

Lambda, The Superintelligence Cloud, is a leader in AI cloud infrastructure serving tens of thousands of customers. Our customers range from AI researchers to enterprises and hyperscalers. Lambda's mission is to make compute as ubiquitous as electricity and give everyone the power of superintelligence. One person, one GPU. If you'd like to build the world's best AI cloud, join us.Note: This position requires presence in our San Jose, San Francisco, or Bellevue office location 4 days per week; Lambda’s designated work from home day is currently Tuesday.

What You’ll Do

• Architect high-performance networking solutions that power cloud platforms, with a focus on ultra-low-latency and high-bandwidth connectivity.
• Define the network topology and architectural patterns for large-scale GPU clusters, storage backends, and multi-tenant environments.
• Evaluate, benchmark, and select next-generation network technologies (e.g., InfiniBand NDR/XDR, RoCE, 400G-1.6T Ethernet, Ultra Ethernet, ESUN, OCI, etc) to meet AI workload requirements.
• Develop and maintain network architecture standards, reference designs, and scalability roadmaps for multi-site and hybrid environments.
• Partner with compute and storage architects to ensure seamless end-to-end data flow and fault tolerance.
• Guide network automation strategies and tooling to enable efficient provisioning, telemetry, and operational visibility.
• Mentor engineers and cross-functional teams on advanced network concepts, troubleshooting, and architectural best practices.

You

• Proven experience (7+ years) architecting high-performance data center networks, preferably for HPC, AI/ML, or large-scale cloud infrastructure.
• Deep expertise with InfiniBand (HDR/NDR) and advanced Ethernet fabrics, including RoCE and RDMA protocols.
• Strong understanding of data center switching architectures, congestion control (PFC, ECN), QoS, and network virtualization technologies such as VXLAN and EVPN.
• Skilled in designing for low-latency and high-throughput data paths, including GPU-to-GPU and storage traffic optimization.
• Expertise with BGP-based fabric design, including eBGP underlays, MP-BGP EVPN, ECMP, route policy, and convergence.
• Experience with SDN, APIs, automation, and network telemetry.
• Strong optical networking competency, including transceiver technologies, fiber types and topologies, link budgets, breakout architectures, and WDM.
• Experience building resilient, fault-tolerant high-performance network architectures with redundancy, failover, and high availability.
• Excellent communication and leadership skills, capable of influencing technical decisions across diverse teams.
• Strong ownership and can do attitude, self-starter who feels comfortable working in ambiguity.

Nice to Have

• Hands-on experience with AI workload profiling, collective communication patterns (e.g., NCCL, MPI), and network tuning for distributed training.
• Familiarity with network automation frameworks and telemetry tools.
• Exposure to DPU/SmartNIC technologies, including NVIDIA Blu