Agentic AI Architect

As the Senior Enterprise Agentic AI Architect you are the technical lead responsible for designing, validating, and evolving the Agentic AI & Process Automation CoE's AI architecture for agentic and automations solutions.

Your mission is to turn business goals and service delivery challenges into robust, production-ready AI architectures that balance performance, cost, safety, and time-to-value. You will define how agent frameworks, enterprise architecture and business platform fit together; you will support pre-sales and solutioning with customers; and you will run an ongoing research and evaluation stream to keep our AI stack state-of-the-art and cost-efficient.

You will work hand in hand with:

• The AI Engineers who build and operate the agents and workflows in production.
• Account teams, solution consultants and service owners in pre-sales and delivery.
• Customer stakeholders, SMEs, admins
• Vendors, platform owners, security, and infrastructure teams to ensure architectures are secure, compliant, and scalable.

This is a senior, hands-on role: you are expected to be able to design diagrams and roadmaps, but also to jump into code, configuration, and PoCs to prove out your designs.

Within SCC, this role sits at the core of the Agentic AI & Process Automation Centre of Excellence, reporting to the Head of Innovation and under the EMEA Digital Workplace organization. The new CoE's purpose will be to:

• Industrialize agentic AI for ITSM in Managed Services for SCC's customers
• Build reusable AI patterns, blueprints, and components that can be applied across multiple customers and sectors.
• Define the target AI architecture (and preferred vendors / platforms) and guide the evolution of existing solutions while constantly evaluating new solutions.
• Govern AI use in production/delivery for customers: safety, reliability, auditability, and cost control.

In the area of Architecture for agentic AI platforms, you will:

• Define end-to-end AI reference architectures for conversational and agentic solutions across chat, voice, and background workflows – covering LLMs, agent runtimes, RAG pipelines, vector stores, tools/APIs, monitoring, and security layers.
• Design multi-agent and tool-using systems where agents plan, call tools, orchestrate automations, and handoff to human agents when necessary.
• Select and combine LLMs and models.
• Shape conversational (chat and voice) architectures that integrate Collaboration platforms like MS Teams, Cisco Webex, other, CCaaS / telephony (including TTS/STT, barge-in, call routing, and human handoff) with agentic logic built on LLMs and tools.
• Design robust RAG and knowledge access patterns.
• Specify integration patterns with enterprise infrastructure systems (identity providers, endpoint management, etc) and business platforms (ServiceNow, MS Dynamics, etc) and other systems and tools involved in the agentic workflows and automations.
• Document architecture artefacts (high-level designs, sequence diagrams, threat models, deployment models, capacity assumptions) that can be consumed by engineering, operations and customer stakeholders.
• Support production operations by design – ensuring high availability, graceful degradation, failover strategies, caching, and fallback behaviours are built into the architecture from the outset.

Within Pre-sales, solution design and customer advisory, you will:

• Engage early with customers in discovery workshops to understand business objectives, service delivery pain points, data landscapes, and constraints (security, compliance, legacy systems).
• Take ownership of the AI architecture elements of proposals, RFP responses and SoWs – translating requirements into a clear technical approach, architecture blueprint, and assumptions.
• Design solution options with clear implications for cost, performance, timelines.
• Create and deliver technical presentations and demos explaining AI concepts in clear business language.
• Lead or support PoCs and pilots – defining success criteria, architecture constraints, datasets, and evaluation methods; then iterating quickly to prove value and feasibility.
• Work closely with sales and account teams to shape AI roadmaps for customers, showing a phased evolution from simple automation to mature agentic processes.

For Research, evaluation and cost/performance optimization, you will:

• Continuously scan the AI landscape (LLMs, multimodal models, vector databases, agent frameworks, evaluation tools, AI safety services) and assess which technologies should be adopted, tested for SCC and its customers.
• Run structured evaluations of models and tools – defining test sets, benchmarks, and scenarios for ITSM and Digital Workplace use cases and comparing quality, latency, robustness and cost across vendors (e.g. Azure OpenAI, Google, open-source models).
• Design cost-efficient architectures that make intelligent use of: prompt and response optimization, context window management, caching, low-rank adaptation / fine-tuning where appropriate, model distillation, and smaller specialized models.
• Collaborate with FinOps / commercial teams to create cost models.
• Define patterns for A/B testing and continuous evaluation of AI behavior in production, using both automated metrics and human-in-the-loop assessments.
• Codify best practices and patterns (e.g. internal playbooks and reference implementations) that make it easier and safer to build new AI use cases on a consistent foundation.

In the area of AI governance, safety, security and compliance, you will:

• Embed safety and guardrails into the architecture, including content filters, policy enforcement layers, access control, PII redaction, and strict constraints on which tools agents can invoke and under what conditions.
• Work closely with security, risk, data protection and compliance teams to ensure architectures meet corporate and regulatory requirements (e.g. GDPR, industry regulations, customer data processing agreements).
• Design logging, observability and audit trails for conversations, tool calls and decisions – enabling investigation, incident response, and model behavior analysis without exposing sensitive data wrongly.
• Establish architectural standards and guardrails for AI across CoE.
• Support incident response and problem management for the AI solutions built, providing deep technical insight into root causes and preventative improvement.

What you bring – experience and skills

• Significant hands-on work with LLMs and Generative AI, building real solutions with modern LLMs.
• Experience architecting agentic and conversational AI solutions – systems calling tools and APIs, orchestrate workflows, and interact with humans across chat and telephony (CcaS).
• Experience with LLM fundamentals: tokens, context windows, temperature, system/instruction prompts, function/tool calling, streaming.
• Practical experience selecting and combining models for different tasks and working with both proprietary and open-source models.
• Experience with at least one major cloud AI stack, ideally Azure.
• Strong hands-on development of AI agents and multi-agent workflows in Python and TypeScript.
• Practical use of agentic frameworks such as LangChain, LangGraph, Semantic Kernel, CrewAI, AutoGen.
• Understanding of planning, tool selection, memory, and human-in-the-loop mechanisms in agent design.
• Experience designing RAG pipelines.
• Familiarity with vector databases.
• Experience integrating with knowledge bases, enterprise infrastructure systems and ITSM platforms.
• Experience designing and troubleshooting REST and GraphQL APIs, webhooks, and messaging/queueing systems.
• Experience implementing monitoring and observability for AI systems – metrics, logs, traces, dashboards – and using them to diagnose issues and drive improvement.
• Familiarity with model lifecycle management concepts: versioning, canary releases, safety evaluations, and rollback strategies.
• Knowledge of AI-specific evaluation approaches and tools.
• Understanding of security best practices for AI: secrets management, network isolation, identity and access management, data encryption, and least-privilege for tools and APIs.
• Awareness of data privacy and regulatory considerations (e.g. GDPR), especially for logs, prompts, responses, and training data.
• Strong coding ability in Python and JavaScript/TypeScript and reviewing others' work.
• Familiarity with modern development practices: Git-based workflows, CI/CD pipelines for AI applications, automated testing, infrastructure-as-code.
• Ability to build small PoCs and internal tools.

As nice-to-have experience that would be highly appreciated, you might also bring:

• Familiarity with ITSM and Service Management processes and tooling (especially ServiceNow) and Contact Centre platforms and processes.
• Exposure to Model Context Protocol (MCP), agent-to-agent protocols or other emerging standards for agent communication and tool ecosystems.
• Experience with Power Platform, Microsoft 365 extensibility, or domain-specific copilots.
• Experience with AI safety testing frameworks or formal risk assessment methodologies.