We are seeking a Data & AI Governance Lead to design, implement, and operationalize enterprise-grade data and model governance for AI initiatives in a regulated environment. This role will ensure that AI, ML, and GenAI solutions are secure, compliant, auditable, and scalable, while enabling innovation across business teams.

The role will work closely with Data Management Office (DMO), Risk, Compliance, Finance, IT, and business stakeholders, translating regulatory expectations into practical, enforceable controls across Azure, Databricks, and AI platforms.

Key Responsibilities

1. Data Governance for AI

• \tDefine and implement enterprise-wide data classification and attribute tagging to support AI usage, including:

◦\tPII / PCI / PHI

◦\tSensitivity tiers

◦\tUsage restrictions and consent

• \tDesign access and entitlement models for AI and agent-based use cases, covering:

◦\tAzure Entra ID integration

◦\tRBAC / ABAC

◦\tPrivileged access management

◦\tBreak-glass procedures

◦\tEnd-to-end audit trails

• \tEstablish the metadata, catalog, and lineage operating model using tools such as Collibra and/or Microsoft Purview, including:

◦\tLineage and traceability for AI pipelines

◦\tPolicies for API-based data acquisition

◦\tGovernance of vector databases and RAG stores

• \tDefine and document cross-border data usage rules, aligned with global and local regulations (e.g., MAS, HKMA, JFSA, GDPR, PDPA).

• \tSet governance boundaries and guidance for Databricks and Azure platforms, including:

◦\tUnity Catalog governance

◦\tLakehouse permissions

◦\tDelta Sharing

◦\tClear "what data can be stored where" principles

2. Model Governance (LLM / ML)

• \tEstablish bank-grade model lifecycle governance, including:

◦\tCentral model inventory

◦\tRisk classification and tiering

◦\tApproval workflows and sign-offs

◦\tModel documentation and validation standards

◦\tHuman-in-the-loop controls

• \tImplement monitoring and observability for ML and LLM models, covering:

◦\tModel drift, bias, and performance

◦\tPrompt and output logging

◦\tToxicity detection and content filtering

◦\tRollback mechanisms and change control

◦\tIntegration with MLflow / Model Registry

• \tDefine and operationalize AI safety processes, including:

◦\tRed teaming and adversarial testing

◦\tPrompt hygiene and secure prompt design

◦\tSensitive data minimization and retention controls

◦\tAI incident response and escalation procedures

3. Ways of Working & Operating Model

• \tDefine the AI governance framework, including:

◦\tStandards, policies, and operating procedures

◦\tClear RACI across Data, Risk, IT, and business teams

• \tPartner with DMO, Risk, Compliance, Finance, and IT to ensure alignment with enterprise governance and regulatory expectations.

• \tTranslate governance principles into practical, implementable controls for:

◦\tAzure cloud services

◦\tDatabricks Lakehouse

◦\tLLM platforms and services

◦\tAI agents used by business teams

• \tAct as a trusted advisor to delivery teams, balancing risk management with innovation enablement.

Candidate Profile (Must Have)

• \tProven experience leading data governance and model governance initiatives in regulated industries, preferably financial services.

• \tHands-on experience with:

◦\tAzure (including Entra ID)

◦\tDatabricks and Unity Catalog

◦\tMLflow / Model Registry

◦\tCollibra and/or Microsoft Purview

◦\tEnterprise IAM and access control models

• \tStrong understanding of regulatory expectations for data usage and model risk management, particularly within APAC jurisdictions.

• \tDemonstrated ability to author policies, standards, and operating procedures, and guide their implementation with both IT and business stakeholders.

• \tExcellent stakeholder management skills, with the ability to influence across senior technical and non-technical audiences.