Senior AI GPU Infrastructure Architect

ECLARO; Costa Mesa (CA); On-site; USD 166,000 - 220,000; Full time

Job summary

ECLARO seeks a Senior AI Infrastructure Engineer in Costa Mesa, CA to lead GPU-scale training infrastructure. You will own the stability of GPU clusters, automate resilience, and optimize NCCL, networking, and scheduling tools to support ML research across multiple teams.

The role demands hands-on hardware management, strong automation, and collaboration with product groups to forecast compute needs and deliver scalable platform capabilities.

Qualifications

• 10+ years in hands-on infrastructure, HPC, or datacenter engineering.
• Experience with H200/B200/B300 GPUs and firmware/driver management.
• Kubernetes required; Run:AI or similar GPU scheduling experience preferred.
• Able to lift/move 50+ lbs and perform physical datacenter work.
• Eligible to obtain and maintain an active U.S. Top Secret clearance.

Responsibilities

• Rack, stack, cable, and bring up GPU compute nodes and validate hardware.
• Build and tune interconnects (NVLink, InfiniBand, RoCE, Spectrum-X) across hundreds of GPUs.
• Integrate high-performance storage (VAST, DDN, Weka) for large datasets.
• Automate cluster deployment end-to-end with infrastructure as code.
• Operate and extend Kubernetes/Run:AI for GPU scheduling and multi-tenant workloads.
• Own fleet health: monitoring, alerting, and rapid triage of hardware and network faults.
• Onboard engineers/researchers to the platform and assist with workload optimization.
• Partner with product teams to translate compute needs into platform capabilities.

Skills

GPU Compute

Kubernetes

Automation

Distributed Training

Tools

Run:AI

NCCL

InfiniBand

DCGM

Lustre

Job description

ECLARO seeks a Senior AI Infrastructure Engineer in Costa Mesa, CA to lead GPU-scale training infrastructure. You will own the stability of GPU clusters, automate resilience, and optimize NCCL, networking, and scheduling tools to support ML research across multiple teams.

The role demands hands-on hardware management, strong automation, and collaboration with product groups to forecast compute needs and deliver scalable platform capabilities.