HPC Infrastructure Architect (AI)

Role Overview

We're looking for an experienced HPC Infrastructure Architect (AI) to design the complete AI Factory stack for a UK-based Service Provider, built on NVIDIA B300 hardware with additional standalone HPE PCAI deployments.

This is a greenfield, design-led engagement covering distributed AI infrastructure, GPU systems, networking, storage, and orchestration- with automation as a foundational element of the build.

You'll take design ownership from environment architecture through to detailed requirements input for the automation workstreams.

What You'll Be Doing:

• Architecting multiple NVIDIA clusters across multiple UK sites for a multi-tenant Service Provider
• Designing the end-to-end AI Factory stack: model training layer, distributed training, model orchestration, data pipelines, RAG, vector DB, and MLOps/ModelOps lifecycle
• Designing multi-cluster resource management and allocation for inferencing and training
• Designing multi-cluster orchestration and control plane (e.g. Rafay) to serve multiple customers/tenants across the UK network of clusters and sites
• Designing for highly available, long-running inferencing and training jobs across multiple clusters.
• Covering GPU pool management and fractional allocation.
• Designing storage and security around the platform and runtime.
• Tuning and optimising for multi-cluster GPU and multi-node scaling.
• Architecting workflows based on personas and use cases.

What We're Looking For:

• Experience designing and building multi-cluster, Service-Provider-grade NVIDIA AI platforms.
• Working knowledge of NVIDIA reference architectures.
• Expertise in resource and bandwidth sizing.
• Working knowledge of Slurm and Run:AI.
• Performance management and tuning across GPU, CPU, storage, and networks
• Knowledge of Infiniband and XDR.
• Ability to provide requirements and design input for automation, covering both build activities and Day 2 operations.
• Ability to design performance, health, and utilisation dashboards, plus telemetry/observability. flows for end-to end and per-tenant management.
• Experience producing High-Level and Low-Level Design documents.
• Comfortable working alongside Red Hat Virtualization (RHV) and container platform. (OpenShift/Kubernetes) teams on cloud-native deployments.
• Experience with Rafay, Ori, or Aarna is a strong plus.
• Experience with integrated storage platforms such as Vast or HPE GreenLake
• Detailed knowledge of the physical installation considerations for this type of deployment
• A flexible, can-do attitude, with strong initiative, the ability to learn fast, and the ability to stay calm under pressure.
• Strong organisation and planning skills to manage multiple concurrent tasks.