sumit10112005/loan-default-predictor ? reverse-engineered prompt

Build me a simple web app called LedgerRisk that lets a user enter borrower details and get a loan default risk prediction. It should show whether the borrower looks risky or not, the estimated default chance, and also place them into a borrower risk group so the result feels more useful than just a yes or no.

I want a clean single page dashboard with a form on one side and results on the other, plus a short explanation of the borrower segment and a few visual touches so it feels polished. Make the app work locally from the browser on localhost, and keep the backend in Python with a small web server. Use the saved model files I already have so the prediction logic matches the trained pipeline, and make sure the input fields are validated before submitting.

If you need to check anything, look up current docs online. Also include a simple README style way to run it locally.