AI Architect, Enterprise Agentic AI Architecture

Job Purpose
We are looking for an experienced AI Architect responsible for architecting enterprise-wide Generative AI and Agentic AI
capabilities across banking systems. The role will define target architecture, integration patterns, standards, reference
implementations and reusable building blocks for full-fledged AI chat assistants, autonomous agents and multi-agent
workflows.
The AI Architect will lead end-to-end architecture for agentic retail banking journeys such as payments, transfers,
servicing, self-service fulfilment and customer assistance, ensuring secure integration with enterprise APIs, middleware,
core banking platforms and customer-facing channels across web and mobile.
This role requires deep hands-on AI engineering capability, strong banking domain understanding, practical delivery
experience with multiple production-grade banking agents, and the ability to collaborate with product, engineering,
infrastructure, cybersecurity, data, governance and enterprise architecture teams to present and align designs through
ARB.
Key Accountabilities
• Define and own enterprise-wide Agentic AI architecture, reference patterns, reusable components and governance
guardrails for banking use cases.
• Architect full-fledged conversational AI and agent platforms for retail banking, including customer assistance,
servicing, payments, transfers and operational self-service journeys.
• Translate business, product and regulatory requirements into scalable, secure, observable and resilient AI solution
architectures.
• Design multi-agent ecosystems with planning, orchestration, tool usage, action execution, memory, context
management, evaluation loops and human-in-the-loop controls.
• Lead the architecture of RAG platforms, knowledge retrieval, semantic indexing, vector databases, grounding
strategies and answer relevancy improvement mechanisms.
• Define token-efficient context engineering patterns to maximize output quality while reducing token utilization,
inference latency and run cost.
• Set architecture standards for model evaluation, agent evaluation, guardrails, auditability, observability,
performance, safety and business outcome measurement.• Architect secure integration with enterprise APIs, backend systems, middleware, channels, data platforms, AI services
and third-party tools using approved enterprise patterns.
• Create Architecture Decision Records, solution architecture documents, integration designs and ARB-ready
presentations to obtain required architecture approvals.
• Collaborate across enterprise architecture, cybersecurity, infrastructure, engineering, product, operations and
business teams to build a sustainable AI ecosystem.
Job Context
Specific Accountability
• Architect and guide implementation of multiple production-grade banking agents, including full-fledged chat
assistants with tool execution, transaction flows and contextual customer support.
• Design agentic workflows for payments, transfers, account/card servicing and self-service fulfilment with strong
controls for authorization, exception handling, recovery and audit trails.
• Design and implement human-in-the-loop patterns for approvals, exception handling, risk-based escalation,
operational review and sensitive customer journeys.
• Define memory architecture including short-term memory, user/session context, long-term knowledge,
personalization boundaries, privacy constraints and secure persistence patterns.
• Architect context management and prompt/context engineering patterns that improve accuracy, reduce latency and
optimize token usage across complex multi-step tasks.
• Design evaluation frameworks for agents, RAG and workflows covering task completion, grounding, faithfulness,
retrieval quality, safety, latency, cost, tool success rate and customer experience.
• Architect enterprise RAG systems including ingestion, redaction, chunking, embeddings, vector search, ranking,
reranking, citation/grounding and quality feedback loops.
• Design and govern tool integration patterns, tool registries, action APIs, orchestration layers and MCP servers for
agent-to-tool and agent-to-system connectivity.
• Apply and guide implementation of agent and AI protocols including MCP, A2A, A2UI and other relevant
interoperability protocols for multi-agent and user-interface integration patterns.
• Define security architecture for AI systems aligned to banking security requirements, zero trust principles, least
privilege, identity-based access, data protection and secure API integration.
• Architect PII redaction, data masking, privacy controls, prompt/input/output filtering, secure logging and audit
mechanisms for regulated banking workloads.
• Design deployment architecture using Azure AI services, Azure OpenAI, AWS AI services, Amazon Bedrock,
Kubernetes, serverless components, observability tools and resilient infrastructure patterns.
• Recommend infrastructure, network, compute, storage, vector database, observability and platform capabilities
required to build and scale enterprise agent architecture.
• Provide hands-on technical direction when required, including building complex proof-of-concepts, reference
implementations, agent workflows and reusable engineering accelerators.
• Mentor AI engineers and architects on agentic architecture patterns, responsible AI, secure design, architecture
trade-offs and engineering best practices.
Added Advantage
environments.
• Prior experience building AI agent ecosystems in regulated financial services or large-scale digital banking
• Hands-on experience with Microsoft Semantic Kernel, AutoGen, LangChain, LangGraph and Microsoft Agent
Framework or equivalent Microsoft-native agent development capabilities.
• Experience establishing platform standards, reusable architecture blueprints and governance models for enterprise
AI adoption.
Health and Safety
AI and Data Roles, Dubai | Page 4 of 20Adheres to all Group H&S policies and Procedure as laid out by the Group.
Frameworks, Boundaries and Decision Making Authority
Functions within the framework and boundaries of Group policies as well as overall organisational and governance
frameworks.
Authorised to take decisions as per the approved authorisation matrix.
Qualifications and Experience
Minimum Qualification
• Bachelor's or Master's degree in Computer Science, Software Engineering, Information Technology, Artificial
Intelligence, Data Science or a related discipline.
• Relevant certifications in cloud architecture, AI engineering, security architecture, enterprise architecture or machine
learning are preferred.
Minimum Experience
• Senior technology professional with around 12+ years of experience across software engineering, architecture,
cloud/platform engineering and enterprise solution delivery.
• Minimum 4+ years of hands-on AI engineering or AI architecture experience, including Generative AI, LLM
applications and Agentic AI solutions.
• Proven experience architecting and delivering multiple banking agents or full-fledged banking chat assistant
capabilities integrated with enterprise systems.
• Strong understanding of banking systems, retail banking journeys, payments, transfers, servicing, customer self-
service, operational controls and regulatory/security considerations.
• Full AI Engineer capability including Python, API integration, microservices, event-driven design, RAG implementation,
model/agent evaluation and cloud-native deployment practices.
• Hands-on experience with agent frameworks and orchestration platforms such as Microsoft Semantic Kernel,
AutoGen, LangChain, LangGraph and similar frameworks.
• Experience architecting MCP servers, tool integration layers, agent-to-agent communication, UI integration patterns
and agent interoperability protocols such as MCP, A2A and A2UI.
• Strong experience with Azure AI services, Azure OpenAI Service, AWS AI services, Amazon Bedrock and related model
deployment/management capabilities.
• Experience designing secure AI systems with zero trust principles, identity and access controls, data protection,
secure API design, PII redaction and privacy-by-design controls.
• Experience presenting solution architecture, trade-off analysis, ADRs and architecture recommendations to ARB or
equivalent architecture governance forums.
• Experience recommending infrastructure architecture for AI platforms including compute, Kubernetes, serverless,
vector databases, observability, monitoring, data pipelines and connectivity.
• Strong capability to collaborate with engineering, product, cybersecurity, infrastructure, operations, data, compliance
and enterprise architecture teams.
Key Technical Skills
• Enterprise Agentic AI architecture, multi-agent systems, autonomous workflows, human-in-the-loop design and full-
fledged chat assistant architecture.
• LLMs, prompt engineering, context engineering, memory design, tool/function calling, agent orchestration,
model/agent evaluation and cost/latency optimization.
• RAG architecture, semantic indexing, embeddings, vector databases, retrieval optimization, reranking, grounding,
answer relevancy and explainability patterns.
• MCP server architecture, tool registries, multi-tool integration, A2A, A2UI, agent interoperability protocols and AI
ecosystem design.
AI and Data Roles, Dubai | Page 5 of 20• Azure AI, Azure OpenAI, AWS AI services, Amazon Bedrock, Kubernetes, serverless, microservices, APIs, event-driven
architecture and observability.
• Security architecture for AI systems including zero trust, PII redaction, data masking, privacy controls, guardrails,
secure logging and auditability.
Behavioural / Leadership Skills
• Strategic architecture thinking with the ability to define enterprise standards, influence platform direction and
simplify complex technical decisions.
• Strong stakeholder communication with the ability to present architecture options, risks, trade-offs and
recommendations to senior leadership and ARB forums.
• Collaborative leadership style with the ability to work across business, product, engineering, cybersecurity, data and
infrastructure teams.
• Hands-on problem-solving mindset, pragmatic decision making, ownership, mentoring capability and commitment to
high-quality secure delivery.
Annexure: Technical and Behavioural Competencies
Technical Competencies
• Enterprise Agentic AI architecture and banking-grade AI ecosystem design.
• Retail banking agent architecture for payments, transfers, servicing and self-service workflows.
• RAG, memory, context engineering, evaluation, tool orchestration and MCP server architecture.
• AI security architecture, zero trust, PII redaction, guardrails, auditability and governance.
• Azure and AWS AI services, cloud-native infrastructure, Kubernetes, serverless and observability for agent platforms.
• Architecture documentation, ADR creation, ARB presentation and cross-team design governance.
Behavioural Competencies
• Strategic thinking, ownership, architecture leadership and continuous improvement.
• Stakeholder management, clear communication, mentoring and cross-functional collaboration.
• Pragmatic decision making, analytical thinking, customer focus and responsible AI mindset.