Security, entitlement and governance

• Design and implement least-privilege access with row-level and column-level security enforced at the data layer, not the BI layer.
• Own and re-run the two-user entitlement test as a mandatory gate before any AI or BI tool goes live against customer data, and after any change to connectors, roles or security configuration.
• Produce the data dictionary and lineage the bank’s audit work needs, reconcile every Finance figure back to source, and build quality tests as code from the first delivery slice.

AI enablement

• Own the Fabric Data Agent (or equivalent) setup that lets approved AI tools query the semantic model via delegated user identity, never a shared service account.
• Design testing that reflects the non-deterministic nature of AI query composition (repeated runs, varied phrasing) while security enforcement remains deterministic.
• Use AI as an accelerator to reverse-engineer and document legacy reporting logic — modelling the business domain from it, never cloning flawed reports.

Regulatory awareness

• Work with the DPO and Compliance on the DPIA before any personal data enters the pipeline; keep AI-facing data to schema, report logic and masked samples by default.
• Design with the EU AI Act’s high-risk obligations in mind for the future AI-assisted transaction-monitoring and payment-screening workstream, without owning compliance sign-off personally.

Experience, Skills & Attributes

• Deep, current Microsoft Fabric expertise: OneLake, Data Factory, Dataflows Gen2, semantic models, Direct Lake, and row-level/column-level security. Comparable depth in Azure Synapse plus Power BI Premium will be considered where Fabric exposure is more recent.
• Breadth across the modern data stack. We are not looking for someone who knows every tool, but your track record should show you have worked hands-on with several: at least one other warehouse technology (for example Databricks, Snowflake, BigQuery or Redshift) and several tools from the wider stack, such as dbt for modelling, a managed ingestion tool like Fivetran or Airbyte, or orchestration tooling.
• Strong SQL / SQL Server background, comfortable owning both pipeline and modelling work rather than only one side.
• Demonstrable ability to reverse-engineer an undocumented legacy reporting system; direct experience of core banking platforms (ideally T24) is a strong advantage.
• Solid understanding of data governance in a regulated environment: lineage, data dictionaries, quality testing as code, and reconciliation to source.

Would be an asset to have…

• Working knowledge of dbt.
• Exposure to AI/LLM-integrated data architectures (semantic layers, agent-based query tools).
• Experience in banking or another regulated financial services environment.