AI Architect

Role Summary

You will partner with the IT Director and internal client teams to translate complex business requirements into scalable, enterprise-grade AI architectures built primarily on Microsoft’s AI platform stack — Azure AI Foundry, Copilot Studio, and the Power Platform — complemented by enterprise agent platforms where they are the right fit. This role owns end-to-end solution design across both pro-code (Python/Azure) and low-code (Power Platform/Copilot Studio) delivery: from client discovery and MVP scoping through architecture governance, integration design, production hardening, and team enablement. You will not be the primary developer, but you will be technically fluent enough to review code, challenge designs, set engineering standards, and guide both pro-code and low-code engineers. You will bring exceptional cloud fundamentals, a platform-first mindset, and the ability to rapidly prototype with AI-assisted development to unblock a high-performing delivery team.

Key Responsibilities

• Client engagement & discovery — Work directly with internal clients to deeply understand their use cases, identify the core problem and success criteria, and translate requirements into a clearly scoped MVP. Act as the technical voice in stakeholder conversations, bridging business need to technical possibility.

• Solution architecture ownership — Design and own end-to-end architectures for AI solutions built on Azure AI Foundry, Copilot Studio, and the Power Platform, complemented by enterprise agent platforms (e.g., Moveworks, Lyzr) where they are the right fit. Produce architecture artefacts (HLD, LLD, ADRs) that guide delivery teams.

• Agent & AI capability design — Lead the design of AI agents and agentic workflows using Copilot Studio and Azure AI Foundry, including MCP-based integration, knowledge grounding (RAG), tool orchestration across connectors and enterprise data, and the evaluation approach that proves each agent does what it claims.

• Data grounding architecture — Design how agents access and ground on enterprise data across Dataverse, SharePoint, Microsoft Fabric/OneLake, SQL, and Azure AI Search or vector stores. Own chunking, retrieval, and indexing strategy, and the judgment on when RAG, fine-tuning, or tool-use is the right pattern.

• Production AI engineering — Own the path from prototype to production: evaluation frameworks (golden datasets, offline and online evaluation, regression testing of agent behaviour), release gating, tracing, and security hardening against prompt injection, over-permissioned tools and MCP servers, data exfiltration, and unsafe content, including red-teaming before go-live.

• Modernization of existing solutions — Assess existing automations and AI assets and define a roadmap to evolve them onto the platform stack in a production-grade, governed, and maintainable way, aligned with enterprise standards.

• Integration architecture — Design how AI capabilities connect to enterprise systems and data using connectors, MCP, and APIs. Ensure clean integration with M365, ServiceNow, ERP, HR platforms, and other enterprise sources.

• Observability & operational excellence — Embed monitoring, tracing, and evaluation into every architecture using platform-native tooling (Azure AI Foundry evaluation and tracing, Copilot Studio analytics, Azure Monitor / Application Insights, Power Platform CoE). Ensure AI solutions are measurable, auditable, and trustworthy at scale.

• Responsible AI, security & compliance — Ensure every solution passes through WTW’s Responsible AI intake and governance processes. Design for data classification, privacy, least-privilege access, and auditability from the outset, and partner with Security, Privacy, and Risk colleagues to meet the obligations of a regulated industry.

• Engineering standards & technical governance — Work hands-on with the delivery team as a trusted design partner. Review code and designs across pro-code and low-code workstreams, including vendor and partner engineers; set standards for authentication, API, CI/CD, and infrastructure-as-code patterns; establish reusable components and reference architectures; and drive ALM and governance across environments.

• AI cost & model economics — Guide model selection, token budgeting, and capacity decisions (e.g., provisioned vs. pay-as-you-go), and embed cost monitoring so AI solutions remain economically sustainable as they scale.

• Technology radar & innovation — Maintain an active awareness of the AI tooling landscape. Evaluate and recommend emerging platforms, frameworks, and patterns that could improve delivery speed, capability, or cost-efficiency for the team.