This role to build the bank’s data platform on the existing on-premise estate, to run the daily controls on it, to put the first ML models into production on the bank’s own transaction data — and to leave behind a person who can do all of it. The last outcome carries equal weight with the first.

Main Duties

• Design and build the reporting data layer: extracts from T24 (Oracle), the card platform, settlement files and the channel databases (SQL Server / PostgreSQL) into a masked, read-only reporting store.
• Build the daily exception controls: failed transactions without a matching credit; authorized-versus-cleared differences; settlement instructions without a confirmed transfer.
• Automate report production starting with the highest-effort manual reports, measuring hours before and after.
• Build and productions the first ML models on the bank’s transaction data, in priority order: anomaly detection on the exception flows (failed transactions, settlement differences), card fraud patterns, then merchant and customer analytics. Models run on-premise (Python, scikit-learn-class stack); every model has a documented owner, monitoring, and a retraining plan before it influences any decision.
• Guide Data Engineer by working together on every pipeline and every model following roadmap: by month 6 Data Engineer modifies pipelines independently; by month 9 Data Engineer runs the daily controls unassisted; by month 12 Data Engineer has built one pipeline end to end alone and retrained one model under supervision. These milestones are in the role’s objectives and reviewed quarterly.
• Mandate that all data pipelines, processes are documented clearly enough for a new team member to run them without prior exposure.

Skill/Requirements

• 5+ years in building data pipelines in production, with at least 2 on banking or payments transaction data.
• Expert SQL on at least two of: Oracle, Microsoft SQL Server, PostgreSQL — the bank runs all three. PL/SQL or T-SQL for stored procedures.
• Expert Python: pandas and the scientific stack for analysis, scheduled jobs and file processing in production, version control as a habit, not a policy.
• Production ML Experience: experience in building, deployed and monitored at least one model on transaction, payments or risk data and can explain what it got wrong, how that was detected, and strategy used to retrain. Fraud/anomaly detection experience weighs most.
• Experience in working with card scheme clearing and settlement files (Visa/Mastercard formats), or with core banking extracts, and can explain a reconciliation break they found.
• Can demonstrate verifiable experience upskilling and mentoring team members.
• Desirable: JasperReports (.jrxml templates, server administration) or a comparable Java reporting stack.