One of the more common questions we have been asked related to both our Series A financingand the development of HadePay.com relates to fighting payment fraud and dealing with AML issues that currently plague blockchain payments.
This is a battle worth fighting, especially for a company that consolidates all payment applications and gateways into a single interface then processes cryptocurrency payments. Due to broad exposure, we believe that HadePay is well positioned to solve complex problems in the fraud prevention, compliance, and user verification space
Fortunately, we are tackling that issue with the work of computer engineer and AI specialist Ajinkya Lahade. Ajinkya is working on HadePay to refine a program he built that predicted fraudulent credit card usage with 99% dependable accuracy. We believe that applying such technology to a service like HadePay.com, which enables so many payment applications and processes cryptocurrency payments to be very important to ensure the service is not abused, and that we can identify, report, and even prevent such threats through the application of technology
See HadePay’s work with Ajinkya on machine learning products with HadePay (learn more)
For those of you who are technical and care to know how we will architect a fraud detection program that will work collectively on more than 50 independent applications, and thereby improve the effectiveness of time-consuming AML compliance programs, first know how HadePay works, then understand that this applied algorithm will work separately from the main HadePay framework. We won’t have to worry about interfacing both, or applying this algorithm to each payment processor.
The above shows Ajinkya’s basic schematic of the credit card fraud detection, a similar model used in a program he previously proved successful. We will use Amazon Sagemaker to automate the process of gathering data and its classification on the basis of tuned algorithms.
We will create the basis of these tuned algorithms from the dataset provided by Kaggle, which has two million data points and will train our model to be deployed on Sagemaker. Then, we will test and refine our model for accuracy.
The transaction data that we believe will be important to gather for the purpose of accuracy will be:
In theory, the model will train on fed data and at the same time continuously test the accuracy of itself. Once a user conflicts the preset parameters like “exceeding transaction frequency” or using the wrong CVV again and again, it will create a fraud report that can be generated into a report for FinCEN or delivered instantly to our admins for review. HadePay can instantly block access and report fraud to more than 50 credit card processors, payment applications, and gateways in extreme instances.
Of course we are still in the early development of this product, and are curious to hear your feedback both positive and negative. We look forward to presenting a crypto fraud prevention service soon that will accurately identify the owner of digital wallets and independently verify ownership before purchases/transactions are complete.