Lead AI Architect - Agentic Solutions

About the role

EY's Client Technology team is looking for a Lead AI Architect in Adelaide to own the solution architecture for a major Oceania agentic initiative. The role modernizes, agentifies and integrates multiple existing products used in GCR professional services delivery for one of Oceania's largest clients, delivering pragmatic, cost-efficient designs that align with global architecture standards, security requirements and the global Tax and Client Technology strategy. It reports to the OCTO Lead Architect and works with product, engineering, data, security and regional teams.

Key responsibilities

• Ensure end-to-end architecture integrity across business process, AI, application, data/knowledge, integration, security and operations, keeping the solution simple yet reusable and production-ready.
• Document architecture decisions, patterns, assumptions, risks and trade-offs, and surface risks early.
• Align the solution with global architecture standards, Tax technology strategy, EY security requirements and approved AI platform direction.
• Clarify the boundary between business self-service, product configuration and engineering-led implementation, and define a repeatable deployment and reuse model.
• Drive AI governance and quality practices, including model selection, prompt and agent evaluation, safety and grounding controls, observability and cost management.
• Ensure secure access to enterprise data and knowledge, with identity, authorization, data privacy, sovereignty, lineage and audit considerations.
• Partner with engineering to keep designs implementation-ready within realistic timelines, costs and platform constraints.
• Produce stakeholder-facing materials such as executive summaries, solution blueprints, decision logs and pattern guidance.
• Mentor technical leads and provide hands-on architectural guidance through design, build, integration, testing, deployment and production readiness.

Required profile

• Working and architectural knowledge of Microsoft Azure cloud architecture, including application, integration, identity, networking, security, monitoring and production operations patterns.
• Knowledge of Azure AI Foundry and Azure OpenAI model consumption, including deployment, prompt engineering, model governance, evaluation and cost-aware usage.
• Knowledge of generative AI and agentic patterns, including agent orchestration, tool calling, skills, workflows, memory, planning and human-in-the-loop controls.
• Knowledge of Retrieval Augmented Generation, including document ingestion, chunking, metadata enrichment, vector indexing, hybrid search, grounding and citation evaluation.
• Knowledge of AI evaluation and quality tooling, including model evaluation, prompt evaluation, RAG assessment, hallucination testing, safety testing and regression testing.