Building an Invoice Predictor with Flask + React

A practical walkthrough of the ML model and app we shipped at HighRadius.

Predicting when an invoice will actually clear is a deceptively hard problem — payment behavior varies wildly by client, region, and season.

We trained a gradient-boosted model on historical clearance data and exposed predictions through a Flask API, with a React dashboard for the finance team to explore confidence bands per client.

The unglamorous part — data cleaning and feature engineering — took far longer than the model itself. That is usually true.

No comments yet