Agentic AI Architect

Agentic AI Architect - J50949

Role & responsibilities

Shift 4 pm - 1 am IST

Here is the job description for the SME1 Agentic AI Architect role:

A little additional color for screening, consistent with what we gave CY: this is not a generic enterprise architect or chatbot role. We need someone who can be the technical authority for Copilot/agentic AI platforms and translate that into production-grade enterprise solutions.

Please probe for:
• End-to-end agent architecture: Can they take a use case from intake through agent design, grounding, actions, guardrails, monitoring, and feedback?
• Platform judgment: Can they explain when to use Copilot Studio versus a custom Azure AI/OpenAI approach, and how Moveworks fits into the landscape?
• Agentic AI depth: Strong understanding of RAG, multi-agent orchestration, planner/executor patterns, context passing, failure handling, and human escalation.
• Enterprise integration: API-first thinking across systems such as ServiceNow, Workday, SAP/JDE, Graph, Power Automate, and other enterprise platformsnot just UI automation.
• Security and governance: Entra ID/RBAC, DLP, data classification, grounding permissions, auditability, responsible AI, hallucination/oversharing controls, and human-in-the-loop where appropriate.
• Production readiness: How they evaluate accuracy and reliability, use test datasets and telemetry, monitor agents, and create feedback loops rather than relying on manual testing alone.
• Real delivery: Ask for examples they personally designed or delivered, including measurable business outcomes. We want builders/architects, not someone whose experience is primarily workshops or high-level enablement.
• Leadership and FDE mindset: They should be able to simplify technical concepts for business stakeholders, mentor engineers, establish reusable patterns/standards, and drive alignment across teams.
•
• A strong SME1 candidate should be able to whiteboard a real enterprise agent solution, explain the architectural trade-offs, and show how they would make it secure, reliable, supportable, and measurable at scale.