Ravelin is a fraud detection company that uses machine learning to pick out fraudsters from a sea of genuine users. But machine learning is only as good as the data that you feed it. Graph network based features are some of the most powerful bits of data we have for this purpose but they are also the hardest to calculate at scale with low latency. In this talk, I'll discuss how we built a unique type of graph database in Go to provide specific data that our machine learning algorithms demand.
YOU MAY ALSO LIKE: