What you'll learn
Learn how to put machine learning fraud detection models into production, using data science algorithms to drive effective models
Karun recently was tasked with an interesting and challenging problem; detecting credit card application fraud at a global financial institution. The client’s current practice of relying on third-party vendors to manage these models was slow, expensive, and time-consuming. Much like hackers, fraudsters are always changing behaviour. To keep up with their tricks, the detection algorithms (models) need to be updated continuously. The longer these models go without updating, the less effective they become, losing money and valuable customers to undetected fraudsters.
Karun’s task was to devise a process to improve their fraud detection with a sophisticated machine learning workflow that empowers data scientists to rapidly and iteratively design and develop new models and put them into production. In this talk, Karun explains how the model was built and how to put machine learning fraud detection models into production, using data science algorithms to drive effective models. Along the way, he explains how a global corporation is creating an extensible platform for more than just application fraud.
A basic understanding of continuous integration and continuous delivery (useful but not required)