Rav'n

Inspiration

We looked into the industry and noticed the lacking knowledge of BVN and its involvement in KYC for the average user as it spans across payment platforms to logistic operations through fingerprint and biometric information. Hence we built Rav'n to include these features and pushed for USSD to include the unbanked population without valid access to the internet.

What it does

E-Commerce and Payment platform with transaction management involving the following features:

Dashboard to display intercepted transaction logs
Fraud Detection Based on Anomaly Detection using TDA
QR Code Payment module

We add a QR code having the merchant contact details, and make it possible to do unbanked payments over USSD in cases where QR doesn’t work so we can promote financial inclusion for members without smartphones.

The QR code function on the buyer’s end works in the following additional areas: Return policy verification of buyer identity which could be integrated with BVN Item confirmation in cases where there needs to clear activity with the buyer and a delivery person, thus preventing fraud

For the fraud detection aspect, we use machine learning models to perform pattern recognition to match customer behavior to certain payment methods. Any out-of-order request either via QR or USSD would require 2-factor verification and the audit log for all transactions would be used to retrain models on new customer activity.

The e-commerce platform would feature items for buyers to purchase with a QR code integration containing details pertaining to the FSI API integration format.

With this, it is possible to add in seamless payment with secure verification patterns over the FSI project

How I built it

The team including I built the platform using NodeJS, Express, Python and SciKit TDA on the machine learning end and the HTML stack on the frontend for showcasing the results of these operations to the end user.

Upon signing up, we came across the workflow for signing up a user using their BVN to include KYC for validation. We base the scope of implementation on a simple merchant application hosting features like transfers for Business To Consumer payment emulation, QR code generation for easy access to user stores etc all based on the BVN concept to identify both user and business identities.

Every transaction is logged on the backend with both users identified with their IDs all backed up by the BVN reference for valid identification.

We generate the QR codes using a QR code generation algorithm encoding user IDs for easy retrieval in the generated image. This allows merchants to advertise securely

On the machine learning end, we used SciKit TDA which bases its implementation on time series data using Wasserstein distance metric. This is used to identify diverging results in normal transaction behaviour using sliding windows in intervals of 2-3 days and more.