Data Science Use Case

What they ask

Data science hiring runs in stages: a technical exam, a first technical interview, a use case to solve, and sometimes a second technical exam. The use case is the deciding step.

Use cases in this area look like, for example:

• You receive a sample of card transactions with customer, merchant category, amount and timestamp. Propose how to flag customers likely to stop using the card within 90 days, including the features you would build, how you would validate, and what you would show the business owner.
• Given loan applications with a small labeled default set, describe a model, how you would handle class imbalance, and which fairness or regulatory checks a bank must run before deployment.

The technical exam and interviews cover statistics, SQL, Python and classic ML (regression, trees, evaluation metrics), and they ask about your own projects in detail.

What they look for

• A structured walk from business question to features to model to metric, with the metric chosen for the business cost (recall for fraud, precision for marketing spend).
• Awareness that a bank cannot ship a black box: explainability, monitoring, and data governance come up.
• Solid SQL; feature engineering questions are often asked as queries.
• Patience: the process is long (weeks to months) and includes medical and security checks after the offer.