Built a banking analytics application on top of a synthetic dataset I generated separately, to practice turning well-structured data into something people could actually use to make decisions, not just a dashboard for its own sake.

The app covers a customer 360 view pulling together accounts, transactions, cards and loans for a single customer, transaction analysis with filtering, fraud and merchant risk monitoring, and loan portfolio performance tracking, with a machine learning component layered on top for predictions.

This is the other half of the same exercise as the dataset project: generating trustworthy data is only useful if something downstream can actually rely on it.

Demo project built to showcase the approach, using synthetic data rather than a live production dataset.