AWS AI Infrastructure Architect

JoB Role: AWS AI Infrastructure Architect in 100% Remote

Role Summary

• The onshore AI Infrastructure Architect is the lead technical authority for the platform and the primary architecture interface to the Customer and AWS.
• This role owns the reference architecture end to end drives key design decisions across compute GPU networking orchestration and DR and governs the single touch site blueprint through its replication across all six edge sites .
• The role leads and aligns onshore and offshore engineering pods.

Key Responsibilities

• Own the end to end reference architecture for the Edge Cloud AIaaS platform AWS landing zone EKS on Outposts edge GPU compute G7 NVIDIA Certified servers and the NVAIE inference stack.
• Design the EKSon Outposts topology choosing local vs extended cluster patterns against disconnect behaviour IPv4 only ALBonly and self managed node constraints.
• Architect the multitenant isolation model vCluster namespace RBAC pertenant KMS network segmentation and RunAI project quota.
• Define the NVIDIA inference architecture NVAIE NIM microservices Dynamo serving DCGM telemetry KEDA autoscaling and the OpenAI compatible API surface.
• Set and govern the orchestration boundary Rafay tenancy lifecycle GitOpscatalog vs RunAI GPU scheduling vs KEDA replica autoscaling avoiding overlapping scaling control.
• Architect HADR warm standby paired sites crossregion EKS controlplane resiliency secretscertconfig replication Outpost re enrolment recovery and RTORPO tiers.
• Define the networking architecture IPv4first addressing designed for later IPv6 introduction Direct Connect Transit Gateway and the Local Gateway responsibility boundary.
• Own the singletouch codified sitebootstrap blueprint Rafay NVAIE RunAI as code that enables replication across all six sites.
• Establish the security architecture image signing Cosign SBOM Syft Wiz scanning gates and policy as code.
• Provide architecture governance review IaC and designs guide engineering pods and manage architecture risks and priced assumptions.

Must Have Skills Experience

• 12 years in IT with 5 years in cloud platform architecture on AWS at production scale.
• Deep AWS Outposts EKS VPC networking Direct Connect Transit Gateway KMS IAM and multi account landing zone Control Tower.
• Strong Kubernetes architecture including multitenancy and handson understanding of EKSon Outposts constraints.
• GPU AI infrastructure architecture NVIDIA GPUs GPU sharing MIG vs fractional GPU and GPU scheduling.
• HADR architecture warm standby multi region resiliency defined RTORPO tiers.
• Infrastructure as Code Terraform and GitOps operating models.
• Strong customer handling skills able to run architecture workshops and design reviews with the Customer and AWS present and defend design decisions to senior stakeholders manage expectations and translate business requirements into technical solutions.
• Stakeholder management and executive communication building trust with the Customer handling escalations and driving consensus across Customer AWS and LTM teams
• AWS Certified Solutions Architect Professional or equivalent depth

Nice to Have Skills

• Rafay or Spectro Cloud RunAI and the NVIDIA Dynamo NIM NVAIE stack
• Telco edge 5G MEC platforms
• SageMaker training runtime data lake Lake Formation Glue
• vCluster and policyascode OPA Kyverno