Principal NVIDIA AI Factory Solutions Architect

Principal AI Factory Solutions ArchitectLocation: Bay Area, CAFTE OnlyMandatory - Nvidia tech stack - AI factory SMEROLE OVERVIEW:Our AI Services and Transformation Unit is seeking a hands-on Technology & Solution Architect who lives at the intersection of NVIDIA’s AI Factory stack and enterprise client delivery. You are not a generalist — you have personally co-designed, configured, and validated AI Factory infrastructure using NVIDIA hardware and software. As GSI partner practitioner, you will work directly with clients and the NVIDIA partner engineering team to architect, co-design, and bring to production end-to-end AI Factory solutions — from GPU cluster topology and NVIDIA DSX validated designs to Agentic AI workload onboarding and token economics benchmarking.WHAT YOU WILL DO:AI Factory Co-Design & Client Advisory• Lead client-facing AI Factory co-design engagements: assess existing infrastructure, define target AI Factory architecture using NVIDIA Enterprise AI Factory Validated Design and Enterprise Reference Architectures, and produce deployment-ready technical blueprints.• Advise clients on full NVIDIA AI Factory stack selection — compute (HGX B200/GB200/GB300, MGX, NVL72), networking (InfiniBand NDR/XDR, Spectrum-X), storage, and the NVIDIA AI software layer — matched to their Agentic AI, Physical AI, or HPC workload profile.• Apply NVIDIA DSX co-designed reference framework to architect modular, gigawatt-scalable AI Factories; use Omniverse DSX digital twin blueprints to simulate and validate designs pre-deployment.• Guide clients on cooling strategy for high-density GPU deployments (40–60+ kW/rack): Direct Liquid Cooling (DLC), immersion cooling, and RDHx — translating physics into procurement specifications and facility requirements.NVIDIA GSI Partner Delivery• Function as unit practitioner-level interface with NVIDIA partner engineering — co-developing client solutions, navigating the GSI validated design process, and maintaining NVIDIA-Certified System configurations across client deployments.• Deploy and configure the NVIDIA AI Enterprise software suite (NIM microservices, NeMo, Nemotron, Dynamo, RAPIDS, Triton, Run:ai, Mission Control) on client AI Factory infrastructure.• Execute Agentic AI workload onboarding onto unified AI Factory platforms: implement NVIDIA AI Blueprint for RAG, configure cuOpt for operational AI, and deploy Kubernetes-native GPU orchestration using Run:ai and NVIDIA GPU/Network Operators.• Run GenAI-Perf and MLPerf benchmarks to validate AI Factory delivery quality; present token throughput (tokens/sec, tokens/watt, tokens/dollar) performance against client SLAs and competitive benchmarks. Technical Solutioning & Pre-Sales Support• Support pre-sales on AI Factory pursuits: build detailed BoMs, architecture diagrams, and NVIDIA stack solution documents for client proposals and RFP responses.• Quantify business value of AI Factory deployments — translate token economics, GPU utilization rates, and inference latency improvements into client ROI models.• Contribute to unit AI Factory IP: delivery accelerators, validated configuration templates, benchmark frameworks, and reusable reference architecture assets.QUALIFICATIONS & EXPERIENCEExperience• 15+ years in technology consulting or infrastructure engineering; 4+ years specifically in AI Data Center, HPC, or GPU infrastructure delivery — hands-on build, configuration, and operate/manage is required.• Direct, verifiable experience working within or alongside NVIDIA’s GSI or partner engineering program — this is non-negotiable.• Proven track record deploying NVIDIA AI Factory components in production: you have racked, cabled, and configured DGX/HGX/MGX systems and validated NVIDIA-Certified configurations — not supervised a team that did.• Experience in client-facing consulting or advisory roles with technical solutioning and proposal ownership.Must-Have DepthComputeHGX B200/GB200, GB300 NVL72, MGX, DGX SuperPOD, RTX PRO ServerNetworkingInfiniBand NDR/XDR, Spectrum-X, NVLink 4/5, ConnectX-8, BlueField-3AI SoftwareNIM, NeMo, Nemotron, Dynamo, RAPIDS, Triton, NVIDIA AI Enterprise, CUDA-XOps & Orch.Mission Control, Run:ai, GPU Operator, NGC, Kubernetes, GenAI- Perf, MLPerfBuild FrameworksDSX Ref Design, Enterprise AI Factory Validated Design, Omniverse DSX Digital TwinData & AgentsAI Blueprint for RAG, cuOpt, MIG, Confidential Compute, NVMe-oF storage• AI Workloads: LLM training/fine-tuning, Agentic AI deployment, inference optimization (FP4/FP8, speculative decoding), token economics benchmarking.Education & Certifications:• BS/MS in Computer Science, Computer Architecture, Electrical Engineering, or equivalent applied engineering background.• NVIDIA NCP-AI or NCP-DS certification strongly preferred. CDCE or equivalent data center credential is a plus.