One oft-overlooked aspect of "Big Data" analytics is data collection: how do you collect massive log data, much of which are semi-structured and/or multi-structured, as close to real time as possible? In our proposed talk, we introduce an open source solution based on MongoDB and Fluentd, the leading open source data collector which, like MongoDB, embraces JSON as its core data format. Our talk features a real world case study by ContextLogic, the company behind the popular personalized shopping platform Wish.com. ContextLogic's lead engineer, Adam Flynn will join me and go over how MongoDB and Fluentd powers their state of the art recommendation engine and A/B testing framework.