Real-time Credit Card Fraud Detector

A description of the images you are showcasing for your project, your process, and the artifacts you created along the way

To help mitigate the number of credit card fraud cases across the world, my team and I decided to develop an application that can determine the likelihood of a credit card transaction being fraudulent. This will help stop fraudulent purchases before they even occur!

The mechanism behind this real-time credit card processing solution is technologies that allows for nearly instant processing of large amounts of incoming data. The front end of the application is simple node.js, while the data retrieval and processing is done with Apache Kafka. The data is finally streamed to Apache Spark, where an intelligent fraud detection algorithm is used to determine the likelihood of fraud, finally the score and decision is passed back to the front end and written to a csv file.

In summary, the project uses Scala, Apache Kafka, Apache Spark, node.js, and instances of shell scripts.

I created the streaming pipeline between the app API and the Scala algorithm, using Apache Kafka to process and manipulate the data real-time. I prepared the data for the algorithm and later stored the results in a file. 

The results of the project for the team was 2nd place overall in the Hackathon, and for me it was some valuable experience in the opensource event-streaming platform Apache Kafka, which I will later on work with at an enterprise-scale through a six-month internship.