To many, "Machine learning" seems like a magical black box that somehow tells us what books we're interested in buying on Amazon, what movies we'd like to watch on Netflix, or what coffee shops we'd like to visit on Foursquare. In this 50 minute session, we'll attempt to demystify some of the ML techniques that are employed in building a recommendation engine for a product catalog, discuss some of the capabilities and limitations of recommender systems, storage and querying solutions for millions of rows and millions of columns, all while avoiding the inevitable rise of SkyNet.