GPU Infrastructure Solution Architect
Remote
Long term contract
Job Profile Summary
The Solution Architect IV is a customer-facing GPU and AI infrastructure architect for US customers. The role turns customers' inference goals into GPU-as-a-Service (GPUaaS) solutions that meet their technical, budget, data-residency and compliance requirements, working with Sales, Solution Engineers, Service Delivery, Support and partners such as NVIDIA and AMD.
It leads discovery, sizes GPU and inference requirements, runs POCs and benchmarks, and presents solutions and business cases to technical and executive stakeholders. The role sits in the GPUaaS team within Rackspace Private Cloud, which sells dedicated and multi-tenant GPU capacity on NVIDIA and AMD Instinct (bare metal, managed Kubernetes, OpenStack and VMware) with managed inference services, and mentors junior Pre-Sales members.
Required Experience
· 8+ years in infrastructure, cloud or solution architecture, including customer-facing pre-sales, consulting or engineering roles.
· GPU infrastructure (mandatory): 3+ years hands-on deploying or operating GPU clusters.
· GPU selling: 1+ years supporting the sale of GPU capacity, GPU cloud or AI infrastructure in a pre-sales or solution role.
· AI data center: Working understanding of AI data center requirements (high-density power, liquid cooling, facility readiness) and how they affect GPU design and timelines.
· Scaling AI workloads: Hands-on experience scaling AI workloads in production, including multi-node GPU deployments and inference autoscaling.
· LLM inference: 1+ years deploying LLM inference in production (such as vLLM, SGLang, TensorRT-LLM or NIM).
· Hands-on coding: Strong, current hands-on coding in Python, plus Bash and Infrastructure as Code (Terraform, Ansible, Helm). Personally builds POCs, benchmark harnesses, sizing tools and inference deployments, not only designs them.
· Hands-on technical skills: Personally deploys and troubleshoots GPU environments: Linux administration, GPU drivers and CUDA/ROCm, containers, Kubernetes (GPU Operator, Helm), standing up and tuning serving engines (vLLM, SGLang), running benchmarks (GenAI-Perf, NCCL tests) and monitoring (DCGM, Prometheus, Grafana).
· Technical Expertise
· GPU infrastructure (working depth expected)
· Accelerators and their sizing characteristics (HBM, memory bandwidth, scale-up domain, power): NVIDIA H100/H200, B200/B300, GB200/GB300 NVL72; AMD Instinct MI300X/MI325X/MI355X.
· Networking, storage and orchestration: NVLink, InfiniBand and RoCEv2; all-flash file systems and GPUDirect Storage; Kubernetes with NVIDIA GPU Operator and AMD ROCm equivalents, MIG and tenant isolation.
· Data center and validation: high-density racks, liquid cooling, burn-in and acceptance testing (DCGM, NCCL/RCCL).
· Inference (working depth expected)
· Serving engines: vLLM, SGLang, TensorRT-LLM, Triton and NIM, AMD ROCm stack; KServe.
· Optimization and sizing: continuous batching, KV-cache management, speculative decoding and quantization (FP8, FP4, INT4); GPU count, type and parallelism to model, traffic and SLA; TTFT, throughput and cost per million tokens.
· Open-weight models (Llama, Qwen, DeepSeek, Mistral, gpt-oss), including MoE models, plus RAG and agentic patterns.
