AI Cluster Architect
Vultr is an active independent cloud-infrastructure company providing global compute, GPU, bare-metal, storage, Kubernetes, and serverless AI inference services.
About Vultr
Vultr provides globally available cloud infrastructure for developers, enterprises, and AI innovators, including Cloud Compute, Cloud GPU, Bare Metal, Cloud Storage, managed Kubernetes, and serverless inference.
Skills
Candidate Availability
Required and preferred rules are kept separate and reflect the wording in the original posting.
About the Role
You will architect large-scale GPU clusters within fixed site power limits, maximizing usable GPU count while leaving headroom for compute, storage and networking. You will model power across the full bill of materials, evaluate fabric topologies, define network scale limits, build configuration templates and capacity models, and document your designs.
Requirements
- 7+ years designing or building large-scale HPC, AI, or hyperscale GPU clusters
- Expert understanding of GPU and accelerator system design including PCIe/NVLink/NVSwitch/ROCm
- Strong familiarity with InfiniBand, RoCE and SpectrumX networking, Clos/dragonfly and large-radix switch design
- Experience modeling power and thermal characteristics of servers, GPUs, NICs, switches, optics and storage
- Ability to design non-blocking or intentionally over/under-subscribed networks
- Ability to analyze vendor SKU-level specifications
- Experience balancing customer requirements for compute, storage and service density
- Strong documentation, communication and collaboration skills
Responsibilities
- Architect large-scale GPU clusters within fixed site power budgets
- Model and validate power consumption across the full cluster bill of materials
- Evaluate tradeoffs across InfiniBand, RoCE, SpectrumX and multi-plane, 2-tier/3-tier, rail-optimized topologies
- Determine network scale limits based on switch radix, link speed, topology and blocking requirements
- Maintain SKU-level power and thermal specifications for GPUs, NICs, switches, DPUs, storage and servers
- Develop power-aware cluster configuration templates and capacity-planning models
- Document architecture, design choices, tradeoffs and operational considerations
- Provide guidance on future-proofing for next-gen hardware
- Collaborate with vendors on novel fabric architectures for 100k+ GPU clusters
Benefits
- 100% company-paid medical, dental and vision insurance premiums
- 401(k) with 100% match up to 4%, immediate vesting
- $2,500 yearly professional development reimbursement
- 11 holidays plus PTO accrual and rollover
- Increased PTO at 3 and 10 year anniversaries
- 1 month paid sabbatical every 5 years
- Annual anniversary bonus
- $500 remote office stipend first year and $400 each following year
- Internet reimbursement up to $75 per month
- Gym membership reimbursement up to $50 per month
- Company paid Wellable subscription