> Markdown version of [/jobs/ext/4768502-ai-infrastructure-architect](https://www.wearedevelopers.com/jobs/ext/4768502-ai-infrastructure-architect). Every page supports `.md` or `Accept: text/markdown`. Links point to the HTML versions so they work for humans too. Agent guide: [/agents.md](https://www.wearedevelopers.com/agents.md). --- # AI Infrastructure Architect - **Company:** Kyndryl, Inc. - **Location:** San Francisco, CA, United States (Remote available) - **Experience:** Expert - **Salary:** $179,760 - $341,520 - **Contract:** Permanent contract - **Skills:** Artificial Intelligence, Cloud Computing, Data Centers, Distributed Systems, Ethernet, InfiniBand, Remote Direct Memory Access, PL-SQL, Systems Architecture, Strategies of Testing, AI Infrastructure, Digital Twin, Graphics Processing Unit (GPU), Cisco - **Published:** November 18, 2026 - **Apply:** https://dejobs.org/x/x/4290EAF6427040ABA6E9A3368F0AE5FB/job/ ## About the Role You're good at what you do and possess the required experience to prove it. However, equally as important - you have a growth mindset; keen to drive your own personal and professional development. You are customer-focused - someone who prioritizes customer success in their work. And finally, you're open and borderless - naturally inclusive in how you work with others., · 10+ years in data center, HPC, cloud, platform, or distributed infrastructure, including architecture leadership. · Depth in at least two of compute, network/fabric, storage, or facilities, with working fluency in the others plus resiliency, security, and lifecycle. · Customer-facing technical leadership, design governance, risk management, and executive communication. Preferred Qualifications · Large-scale HPC or supercomputing architecture. · OEM GPU platforms, RDMA/RoCEv2 fabrics, parallel storage, liquid cooling, or cloud GPU architecture. · Cisco, Dell, HPE, Lenovo, Supermicro or equivalent validated-system architecture. · NVIDIA infrastructure certification (or equivalent demonstrated delivery expertise) and cloud or data center architecture certifications. Required Skills · DGX/HGX, GB200/NVL-class or current NVIDIA rack-scale platforms. · InfiniBand, Spectrum-X, BlueField, ConnectX, UFM, NVLink/NVSwitch, BCM, Mission Control. · Digital-twin and simulation tooling such as NVIDIA Air and Omniverse DSX blueprints. ## Description Accountable for the end-to-end architecture and technical decisions that keep the AI factory deliverable, supportable, and acceptable to the customer. Provides design authority across facilities, compute, fabric, storage, platform, security, and lifecycle, and ensures risks are resolved before equipment arrives., · Lead discovery and convert workload, scale, performance, availability, security, compliance, data-locality, and operating-model requirements into an executable architecture. · Select the appropriate NVIDIA reference architecture, OEM validated design, or supported equivalent pattern and document deviations, risks, assumptions, and decision rationale. · Design compute, GPU, accelerator interconnect (for example NVLink/NVSwitch), InfiniBand or RoCE/Ethernet fabrics (for example Spectrum-X), storage, management, out-of-band, and service-network planes. · Define rack elevation, power density, redundancy, cooling or liquid-loop requirements, floor loading, containment, service clearances, failure domains, and capacity headroom with facilities teams. · Produce and approve HLD, LLD, BOM, topology, IP plan, port and cable matrix, compatibility matrix, security requirements, implementation guidance, test strategy, and acceptance criteria. · Chair architecture and design reviews; resolve cross-domain decisions with customer, OEM, network, storage, platform, facilities, security, and operations stakeholders. · Define expansion units, scale boundaries, lifecycle strategy, upgrade paths, rollback principles, and Day 2 operating model. · Mentor engineers and retain design authority through build, commissioning, validation, production handoff, and optimization.