Inspiration
Online stores need to make quick payment decisions while protecting customers and merchants. We wanted to explore how a dashboard could make unusual payment activity easier to spot and explain.
What Safeora AI does
Safeora AI monitors demo transactions and assigns each a risk score. It combines visible risk rules—such as an unusually large purchase, a new device, a location mismatch, or repeated failed attempts—with an Isolation Forest anomaly model trained on synthetic demo data. The dashboard shows the score, status, and reasons behind the assessment.
It also includes a demo checkout connected to the PayPal Sandbox. Test payments can be captured and then shown in the transaction dashboard. Sandbox payments use test accounts and do not move real money.
How we built it
We built the app with Python and Flask, a browser dashboard using HTML, CSS, and JavaScript, scikit-learn for anomaly detection, and PayPal’s Orders API for Sandbox checkout.
Challenges and what we learned
Connecting the checkout required separating the Sandbox buyer and seller accounts and carefully checking the payment flow. We learned how to connect a payment API to a web app, combine machine-learning anomaly detection with clear rules, and show why a transaction received its score.
Safeora AI is a hackathon demo using synthetic data. Its scores are for demonstration and are not real payment decisions.