Principal Architect AI Infrastructure

World Wide Technology is looking for a Principal Architect – AI Infrastructure to join our Strategic Resourcing practice. This is a contract role supporting one of WWT’s enterprise customers. The selected candidate will be employed by one of WWT’s Empaneled partner. This role offers the opportunity to work alongside highly motivated individuals on high-performance teams delivering impactful network solutions in complex enterprise environments. Role Title: Principal Architect – AI Infrastructure Location: Sydney / Melbourne / Perth/Adelaide Employment Type: Contract About the Role The Principal Architect – AI Infrastructure is the design authority for the “Compute-Network-Storage” triad and the governing technical owner of the physical AI factory. Where the Domain Architects own the engineering of each layer and the offshore squads own the build, you own the reference architecture, the standards against which every cluster is judged, and the technical coherence of the triad as a single system. You are the technical judiciary for the infrastructure stream: you adjudicate conflicting engineering approaches, approve deviations from the “Gold Standard”, and delegate the L3 fix to the responsible Domain Architect. As a System Integrator, we do not simply manage a static cloud; we design and deliver bespoke, high-scale AI factories for the world’s leading enterprises. You are a consultative and governing authority, not an engineer-of-record. You reason fluently across silicon, fabric, and data platform – as comfortable interrogating a rail-optimised topology or a parallel filesystem sizing as defending a Bill of Materials against budget – but you draw hands-on depth from the Domain Architect pool and the offshore Senior HPC Engineers rather than owning the keyboard yourself. You own the “Gold Standard” reference architectures (NVIDIA SuperPOD, NVIDIA BasePOD, Cisco AI Factory) for NVIDIA Cloud Provider (NCP) and private enterprise AI cloud deployments, and the deviation gates that protect them. On infrastructure-led engagements – the “Turnkey AI Data Center” – you are the Prime. You own the final acceptance criteria and the Architecture Review Board (ARB) approval, defend the solution before the client’s Design Authority, and direct the Platforms, Solutions, and Facilities Principals as internal stakeholders and suppliers to the build. You are the originating technical owner who converts the sanctioned Horizon 2 and Horizon 3 programs handed down by the Enterprise Architect into a buildable, governed reference design. This position operates with a 60/40 split between Technical Authority & Delivery Governance (60%) and Pre-Sales, Commercial & Practice Development (40%). The 60% is governance, design authority, and delivery oversight exercised through the Domain Architects and squads – not personal hands-on implementation. Key Responsibilities 1. Technical Authority & Delivery Governance (60%) Reference Architecture Ownership:Own and maintain the “Gold Standard” reference architectures for the triad (NVIDIA SuperPOD, NVIDIA BasePOD, Cisco AI Factory), keeping them repeatable, scalable, and commercially defensible across NCP and private enterprise AI cloud builds.Chair the internal Architecture Review Board (ARB), approving or denying deviations from the reference architecture and owning the record of why each deviation was permitted.Ensure HLD/LLD coherence across Compute, Network, and Storage so the three layers integrate as one fabric: NUMA/PCIe affinity aligned to rail-optimised topology, storage clients aligned to the RDMA fabric, and validation targets (NCCL/HPL, IOR/FIO) consistent across the triad.Design Authority & Client Engagement:Act as the Design Authority on infrastructure-led engagements: defend the architecture before the client’s Design Authority / ARB and own the final technical acceptance criteria as Prime.Approve reference-architecture deviations affecting the global fabric, security posture, or supportability, and hold logical-design authority for changes the Field Solutions Engineer and Domain Architects escalate from site.Coach the Domain Architects on “the Story”: translating engineering decisions into a narrative a client CTO and a procurement function will both sign.Domain Architect Leadership:Line-manage and technically develop the Compute, Network, and Storage Domain Architects, owning their technical bench health, skills plan, and succession.Arbitrate cross-domain technical disputes within the triad and across streams (for example, host-networking ownership between Compute and Network, or the RDMA fabric between Storage and Network), preventing “design-by-committee” stalls.Govern the layered delivery model: Domain Architects direct the offshore Senior HPC Engineers, who direct the HPC Engineers, holding the architects accountable for HLD/LLD quality and first-pass validation success.Delivery & Commercial Governance:Validate the consolidated infrastructure Bill of Materials against budget and against the NVIDIA HCL / OEM compatibility matrices before commitment, and own margin on infrastructure delivery.Govern Labour Estimate (LOE) and Statement of Work quality produced by the Domain Architects, holding estimation variance within practice tolerance.Approve PoC GPU/cloud burn within delegated authority, escalating spend above the agreed per-PoC threshold to the Head of AI.Cross-Stream & Overlay Integration:Define the interfaces between the triad and the adjacent Facilities (Layer 0/1 power, cooling, structured cabling) and Edge Infrastructure domains owned by the peer Principal, reconciling the physical envelope with the logical design.Coordinate with the Platforms and Solutions Principals on the “Day 2” handover (orchestration, identity, storage backends) and with the Prime Principal Architect overlay for North Asia, where the Field Solutions Engineer holds site authority.2. Pre-Sales, Commercial & Practice Development (40%) Opportunity Shaping & Prime Selection:Receive qualified Horizon 2 and Horizon 3 infrastructure programs from the Enterprise Architect and shape them into buildable programs with an explicit reference architecture, charter, and acceptance criteria.Produce the Program Charter and the technical sections of the proposal, assuming Prime on engagements where physical capability (power/compute) is the primary value driver.Commercial & Partner Alliance:Own the Partner Selection Matrix for the triad (compute OEM, fabric, and storage vendors) and the technical side of the NVIDIA and OEM alliance relationships.Defend the architecture commercially: build-versus-buy, BoM versus budget, and the utilisation and depreciation consequences of the GPU hardware cadence, with the Technical Account Manager and Enterprise Architect.Standards, Repeatability & Practice Capability:Convert bespoke pre-sales designs into reusable reference architectures and Delivery playbooks, industrialising custom work into standard patterns the offshore factory can execute.Define the Infrastructure squad skills plan and validate resource fit against engagement complexity (C1 PoC through C5 sovereign).Horizon 3 Roadmap:Own the forward infrastructure roadmap, evaluating emerging compute paradigms against the established triad: novel silicon (neuromorphic, LPU/Groq disaggregated inference), scale-across fabrics (Metro-Xacross data-centre halls), and the path from proprietary InfiniBand collectives to open Ethernet (UEC) standards.Govern data-residency and delivery-model fit against the regional tier framework (Tier 1/2 jurisdictions, commercial-versus-sovereign engagement), ensuring the reference architecture supports the required staffing and execution model.Technical Competencies Essential Skills HPC Mastery (Cross-Domain):Demonstrated authority across Compute, Network, and Storage, with the ability to reason about their interaction as a single system; depth in at least one layer and credible breadth across the other two. This is not a single-domain specialist role.NVIDIA Reference Architectures:Command of SuperPOD and BasePOD reference designs; NVL72/DGX/HGX/MGX compute; Quantum InfiniBand and Spectrum-X Ethernet fabrics; and high-performance parallel storage (VAST, WEKA, DDN, Pure).Working knowledge of NCP and Cisco AI Factory build models.Architecture Governance:ARB facilitation, reference-architecture and deviation control, and BoM/HCL governance, holding a design defensible under a client Design Authority and an internal margin review at once.Commercial Literacy:LOE/SOW construction and review, infrastructure delivery margin, BoM-versus-budget, and TCO/utilisation reasoning sufficient to underwrite a build commercially.Architectural Leadership:Line management and technical coaching of senior architects, and governance of multi-squad delivery through a layered architect / senior engineer / engineer model.Linux & IaC Fluency:Sufficient depth in Linux systems engineering and Infrastructure as Code (Ansible, Python, Terraform) to govern automation quality and review the squads without owning implementation.Desirable Experience Industry Background: Prior tenure as a Domain Architect or lead infrastructure architect within a System Integrator (SI) or MSP, with at least one greenfield AI factory delivered end to end.Platforms & Solutions Adjacency: Working understanding of the “Day 2” stack (Kubernetes / Red Hat OpenShift, Rafay, Run:AI) and the MLOps/LLMOps handover, sufficient to manage the cross-Principal interface.Multi-Vendor Exposure: Cisco AI Factory and hyperscaler (Azure/AWS/GCP/OCI) AI infrastructure design alongside the NVIDIA-native stack.Novel Compute: Familiarity with the emerging-silicon roadmap (neuromorphic, LPU, quantum infrastructure) and its implications for the reference architecture.APAC Regulatory & Sovereignty: Awareness of data-residency, export-control, and sovereign-delivery constraints across APAC Tier 1/2 jurisdictions.Certifications (Preferred) NVIDIA-Certified Professional: AI Infrastructure (NCP-AII)NVIDIA-Certified Professional: AI Networking (NCP-AIN)NVIDIA-Certified Associate: AI Infrastructure and Operations (NCA-AIIO)TOGAF or equivalent enterprise-architecture certification (advantageous for the cross-stream governance remit)Success Metrics (KPIs) Architecture Adoption: Infrastructure engagements run against an approved reference architecture, with deviation rate controlled and every deviation recorded with rationale.First-Pass Delivery Quality: Delivered clusters pass NVIDIA validation (NCCL/HPL) and storage acceptance (IOR/FIO) on first handover across the triad.Design Defensibility: No architecture reversed on a feasibility ground that should have been identified at ARB; designs validated as sound by the client Design Authority.Commercial Control: Infrastructure delivery margin held to target, with BoM-versus-budget and LOE estimation variance within practice tolerance.Practice Capability: Domain Architect bench depth and skills plan maintained, and bespoke designs industrialised into reusable playbooks.Pipeline Contribution: Horizon 2 and Horizon 3 infrastructure programs shaped and won as Prime. We strive to create an environment where all employees are empowered to succeed based on their skills, performance, and dedication. Our goal is to cultivate a culture of belonging that encourages innovation, collaboration, and respect for all team members, ensuring that WWT remains a great place to work for All! Equal Opportunity Employer