Senior Solutions Engineer, GPU & AI Infrastructure | GPU Neo-cloud | UK (Remote-first)
A high-performance GPU Neo-cloud provider is hiring a Senior Solutions Engineer to lead the technical side of its largest AI and HPC customer engagements. The business is competing directly with the hyperscalers on cost and complexity. It runs the latest NVIDIA Blackwell platforms, including B300 and GB300 NVL rack-scale systems, and has a clear mandate to win large, high-value AI infrastructure deals.
You will be the lead technical architect on those deals. That means taking a customer’s ambitions for training and inferencing foundation models and turning them into production-ready GPU supercomputing designs, from the first workshop to PoC sign-off.
What’s in it for you?
- Own the architecture of frontier infrastructure: B300 and GB300 NVL platforms, InfiniBand and Spectrum-X fabrics, and 100kW+ liquid-cooled racks
- Join a Neo-cloud at a stage where it is winning large AI infrastructure deals, with direct access to the latest NVIDIA GPU clusters
- A 4-day working week, uncapped holiday and remote-first working, with real autonomy inside a collaborative team of specialists
- Work across cutting-edge hardware, executive-level customer conversations and hands-on validation, all in one role
Responsibilities
- Lead the technical side of deals alongside sales: run architecture workshops with CTOs, Chief AI Officers and ML engineers, and own technical responses to complex RFPs and RFIs
- Write enterprise-grade HLD and LLD documentation for large-scale GPU clusters, with peer-reviewed designs delivered within deal timelines
- Produce detailed Bills of Materials covering compute, NVLink switching, fabrics, optics, cabling, liquid and air cooling, power distribution and parallel storage
- Architect scale-up (NVLink/NVSwitch) and scale-out (Fat-Tree, Rail-Optimised) topologies across InfiniBand (NDR/X800) and RoCEv2 (Spectrum-X), including lossless Ethernet tuning (PFC, ECN, Adaptive Routing)
- Design bare-metal (Slurm, OpenMPI) and cloud-native (Kubernetes, GPU Operator, Network Operator, Run:ai, Kubeflow) multi-tenant architectures integrated with GPUDirect Storage and high-performance file systems such as VAST Data
- Design and oversee PoCs: benchmark with NCCL tests, GPUDirect RDMA, MLPerf and Megatron-LM, resolve congestion, routing and thermal/power issues, and feed what you learn back to product and platform engineering
Skills Required
- 5+ years in Solutions Architecture, Systems Engineering or Technical Pre-Sales within HPC, AI infrastructure or high-performance cloud
- Hands-on experience with NVIDIA HGX/DGX and Blackwell platforms
- Expert-level InfiniBand and RoCEv2 networking
- Proven authorship of HLDs, LLDs, rack diagrams and BOMs for GPU clusters
- Kubernetes and Slurm based GPU cluster deployment
- UK based
Strong Differentiators
- NVIDIA certifications: NCP-AII, NCP-AIN, NCP-IB, or AI Workload Deployment & Cloud Native
- Direct experience with GB200 NVL72/36, GB300 NVL or Quantum-X800 / Spectrum-4
- Liquid cooling knowledge (direct-to-chip, CDU) and design experience with 100kW+ per rack power constraints
- PyTorch/NCCL tuning and diagnosing hardware-software bottlenecks in distributed training and inference
- Confident presenting to executives and explaining complex network and hardware trade-offs clearly
- Degree in Computer Science, Electrical or Systems Engineering, or equivalent practical experience