AI Platform Architect

Job Description & Summary

Design and evolve the reusable platform capabilities required to build deploy secure observe and scale AI products across hybrid environments.

Define reference platforms environment topology network patterns identity secrets and access controls.

Design model access compute storage vector services orchestration gateways and shared platform services.

Support cloud sovereign edge and on-premises deployment choices where client constraints require them.

Define platform non-functional requirements for availability performance resilience scalability and cost.

Create reusable landing patterns templates and platform guardrails for delivery squads.

Partner with the MLOps Engineer on environment automation observability and release management.

8 years in cloud platforms infrastructure platform engineering or architecture.

Deep understanding of containers Kubernetes networking identity automation data services and AI runtimes.

Experience with at least one major cloud platform and hybrid integration patterns.

Strong Infrastructure as Code DevSecOps and platform-governance orientation.

Deep experience with an enterprise AI platform such as Azure AI Foundry and Azure OpenAI AWS Bedrock Google Vertex AI Databricks or an equivalent platform plus the ability to integrate alternative model providers.

Hands-on knowledge of containers and Kubernetes platforms such as AKS or OpenShift; Infrastructure as Code using Terraform Bicep or equivalent; and GitHub Actions or Azure DevOps pipelines.

Experience exposing and securing model agent and tool services through API gateways private endpoints service identities secrets platforms and policy enforcement for example Microsoft Entra ID API Management and Key Vault.

Ability to design hybrid inference patterns using managed endpoints and self-hosted runtimes such as vLLM Hugging Face tooling NVIDIA inference components Ollama or equivalent technologies where appropriate.

Knowledge of vector search and state services such as Azure AI Search PostgreSQL with pgvector Elasticsearch Pinecone Weaviate Milvus Redis Cosmos DB or equivalent.

Experience integrating platform telemetry with OpenTelemetry and enterprise monitoring stacks and designing for model or framework portability rather than unnecessary vendor lock-in.

Platform readiness and reliability

Time required to onboard new use cases

Reuse of platform components

Cost performance and capacity transparency

Compliance with approved architecture guardrails

Other members of the AI Transformation & Agentic Systems Practice

PwC sector functional cloud cyber risk Responsible AI and change specialists

Client business owners product owners technology teams and operational users

Technology alliance and implementation partners where relevant

Support proposals client workshops and market development appropriate to seniority.

Contribute reusable methods patterns code assets and lessons learned.

Coach colleagues and participate in the capabilitys continuous learning agenda.

Uphold PwC quality independence confidentiality and risk-management requirements.

#LI-BS1 #LI-Hybrid