AI Cluster Architect

About this role

Design and refine large-scale GPU cluster architectures within fixed power, thermal, and facility limits. Optimize GPU capacity while balancing compute, storage, networking, and customer performance requirements.

About the company

Vultr provides cloud compute, storage, and GPU infrastructure across global data centers. It serves developers and businesses deploying applications on flexible cloud resources.

Stack

What you'll do

• Architect GPU clusters to maximize GPU density within site power budgets.
• Model power and thermal use across the full cluster bill of materials.
• Evaluate networking fabrics, topologies, switch limits, and blocking tradeoffs.
• Develop reusable power-aware configuration templates and capacity-planning models.
• Document architecture decisions and lifecycle considerations.
• Collaborate with vendors on fabric designs for deployments of 100,000+ GPUs.

What we're looking for

• 7+ years designing or building large-scale HPC, AI, or hyperscale GPU clusters.
• Expertise in GPU system topology, interconnects, and NIC-to-GPU affinity.
• Strong familiarity with InfiniBand, RoCE, Spectrum-X, and large-scale network topologies.
• Experience modeling power and thermal characteristics across servers, networking, and storage.
• Ability to design non-blocking or intentionally oversubscribed networks and assess workload impact.
• Experience analyzing vendor SKU specifications and balancing compute, storage, and service-density requirements.
• Strong documentation, communication, and collaboration skills.

Benefits

• Company-paid employee medical, dental, and vision insurance premiums.
• 401(k) match of 100% up to 4%, with immediate vesting.
• $2,500 annual professional development reimbursement.
• 11 holidays and accruing PTO with rollover.
• Additional PTO at service milestones and a paid sabbatical every five years.
• Remote office stipend and internet reimbursement.
• Gym membership reimbursement and a company-paid Wellable subscription.