Director, AI Solution Architect

SMBC Asia Pacific has established a dedicated Artificial Intelligence Cross-Functional Team (AI CFT) to accelerate productivity, operational efficiency, risk management, and client experience through AI-enabled automation, agentic AI, and intelligent process optimisation. The Senior AI Solution Architect is the lead solution architect responsible for defining the architecture, design principles, engineering frameworks, and governance standards for enterprise AI solutions delivered across Asia Pacific. The role translates prioritised business use cases into secure, scalable, reusable, and production-grade solutions across the Microsoft ecosystem, including Microsoft Copilot Studio, Microsoft Foundry, GitHub Copilot, Power Platform, Dataverse, Databricks AI, and related Azure services and third party tools on Microsoft ecosystem such as Claude Code. Working with the AI CFT Technology Solutioning & Delivery Lead, business stakeholders, enterprise architecture, cybersecurity, risk, compliance, data, infrastructure, and engineering teams, the incumbent will establish reference architectures and delivery guardrails for AI agents, both hard-code and low-code solutions, Retrieval-Augmented Generation (RAG), and AI-assisted software development. The role will balance rapid innovation with responsible AI, secure software development, operational resilience, and regulatory requirements across APAC jurisdictions. Responsibilites: Enterprise AI Architecture and Design Authority • Define end-to-end solution architectures for enterprise AI use cases leveraging Microsoft Copilot Studio, Microsoft Foundry, GitHub Copilot, Power Platform, Dataverse, Databricks AI, Azure OpenAI, and related Azure services and third party tools on Microsoft ecosystem. • Establish and enforce enterprise AI design principles covering security and privacy by design, responsible AI, human oversight, least privilege, reusability, modularity, scalability, resilience, explainability, traceability, auditability, and technology portability. • Develop target-state architectures, reference architectures, reusable solution patterns, architecture decision records, non-functional requirements, and implementation roadmaps aligned with SMBC enterprise standards. • Lead architecture reviews, solution design workshops, technical design authorities, and governance forums; identify architectural risks, dependencies, trade-offs, and remediation actions. • Define build-versus-buy and technology-selection recommendations, ensuring solutions are maintainable, cost-effective, supportable, and suitable for regulated banking environments. AI Delivery Framework, Governance and Responsible AI • Define the end-to-end AI solution lifecycle framework covering use-case assessment, risk classification, architecture and security review, data assessment, model and prompt selection, development, evaluation, deployment, monitoring, incident management, periodic review, and retirement. • Design control measures for prompt injection, data leakage, harmful or unsupported output, hallucination, model drift, excessive agency, inappropriate tool use, and unauthorised transactions. • Define human-in-the-loop controls and approval checkpoints for sensitive, high-risk, financial, customer-impacting, or irreversible activities, with clear accountability across business and technology owners. AI Platform Architecture • Architect enterprise-grade solutions using Power Apps, Power Automate, Dataverse, AI Builder, custom connectors, Copilot Studio, Microsoft Foundry, and Databricks AI. • Define the AI solutioning design and governance framework with AI platform such as Power Platform, Microsoft Foundry, Databricks etc., including environment strategy, development/test/production segregation, maker permissions, role-based access, Data Loss Prevention policies, connector and plugin governance, knowledge-source onboarding, authentication, publishing approvals, and support ownership. • Establish Application Lifecycle Management and CI/CD standards for AI Platform solutions, including managed solutions, source control, automated deployment, versioning, rollback, testing, release approvals, and production monitoring. • Define reusable patterns for custom copilots, multi-agent orchestration, knowledge integration, generative experiences, conversation design, human hand-off, transactional actions, exception handling, and human approvals. • Set standards for system instructions, prompt design, grounding, topic management, response validation, citations, fallback behaviour, monitoring, capacity, licensing, and cost controls. GitHub Copilot Development Framework and Governance • Define and govern the enterprise adoption framework for GitHub Copilot across software engineering teams, repositories, programming languages, and development environments in a scalable way to other platform such as Claude Code. • Establish approved use cases and guardrails for AI-assisted code generation, code explanation, refactoring, unit-test generation, documentation, pull-request support, and secure coding. • Define policies for access provisioning, licence assignment, eligible repositories, confidential or regulated information, secure prompt practices, public-code matching, intellectual-property risk, restricted use cases, monitoring, and periodic access review. • Integrate GitHub Copilot into the enterprise SDLC and DevSecOps framework, ensuring AI-generated code remains subject to developer accountability, peer review, pull-request approval, static application security testing, software composition analysis, secret scanning, dependency scanning, and required testing. • Develop reusable repository instructions, coding standards, prompt patterns, developer playbooks, and adoption metrics covering quality, productivity, security findings, developer experience, and realised value. Agentic AI, RAG and AI Engineering • Design enterprise AI agent frameworks using Microsoft Copilot Studio, Microsoft Foundry, Semantic Kernel, LangGraph, LangChain, AutoGen, or equivalent technologies. • Architect RAG and GraphRAG solutions, including content ingestion, chunking, indexing, embeddings, vector search, metadata and access-control filtering, retrieval, reranking, grounding, citations, evaluation, and monitoring. • Design agent identity, memory, planning, orchestration, tool usage, permissions, transactional boundaries, failure handling, observability, and human-in-the-loop controls. • Define patterns for conversational AI, workflow agents, and controlled autonomous agents, ensuring agent actions are secure, explainable, auditable, and aligned with business authority levels. • Establish evaluation datasets, automated test harnesses, tracing, feedback loops, and continuous-improvement mechanisms for AI models, prompts, agents, and knowledge retrieval. Enterprise Integration and Non-Functional Architecture • Design secure integrations with APIs, microservices, databases, data platforms, CRM and ERP systems, banking applications, collaboration platforms, document-management systems, and enterprise knowledge repositories. • Define API-led, event-driven, and microservices-based interaction patterns between AI agents and enterprise systems, including authentication, authorisation, managed identity, secrets management, encryption, network security, and API management. • Ensure solution designs address performance, scalability, availability, resilience, disaster recovery, observability, supportability, data residency, retention, and operational controls. • Align solutions with enterprise architecture, cloud, data, integration, cybersecurity, technology risk, model risk, and regulatory standards. Requirements: • 12+ years of experience in technology, solution architecture, enterprise architecture, application engineering, or digital transformation, preferably within banking, financial services, or another regulated industry. • 5+ years of exp