Gravity’s mission is to personalize the internet by generating interest graphs that help websites deliver customized content to every site visitor. A single publisher may post thousands of articles in a single day. How should it decide which pieces of content to feature on the homepage? Many online publishers today show different pieces of content to different segments of users who are identified based on similar behaviors, preferences and interests. But Gravity is helping these publishers optimize their content delivery even further by offering totally customized content to every individual. Gravity started with a MySQL infrastructure, as many start-ups do. It’s easy to deploy, scales horizontally, and it works. “But at a certain point,” said Benedetto, “we realized that having 500 to 1,000 MySQL boxes running autonomously and building sharding code to keep track of all that data was becoming overwhelming.” Gravity collects and processes more than 10,000 data points every second. All of the data Gravity collects is loaded into HDFS, where Hadoop processes run. Gravity’s back-end ontology is comprised of tens of millions of nodes that solely contain interest data today. The number of nodes capturing data that connects people to their interests is in the billions. With Gravity’s Cloudera-powered environment, it’s able to target tens of millions of content pieces to online users every day. Most publishers or content targeting systems require thousands of data points before they can start personalizing content. But Gravity’s dynamic, real-time system starts personalizing the experience after just a few clicks.