At Sift Science, we protect online businesses from fraud and abuse. We do this by processing lots of data sent to us by our customers, building machine learned models that are updated online in real-time, and surfacing risk scores and decisions to our customers.
We have many different analytics workloads which are served by Apache Crunch, Spark, and Phoenix.
For very fast ad-hoc analysis of our real-time data we use Druid! We presented our experience setting up and using Druid along with specific issues that we resolved at a recent Druid meet-up hosted here at Sift Science. Below is a recording of our presentation. Enjoy!