Effectively scaling the world’s largest social platform is a unique engineering challenge. Over the past year Facebook has added hundreds of millions users and launched a torrent of new features at a rate much faster than our engineering organization has grown. Building abstractions around the constraints imposed by that scaling is unique design challenge.

Throughout this growth, we've also substantially redesigned our internal programming abstractions to make our engineering organization more productive and less error-prone. We have developed fbobjects -- a schema-less data storage engine; Preparables -- a pattern for orchestrating data-fetching amongst disparate components; and Ents -- an object model that allows us to declaratively fetch and manipulate privacy-aware views of the Facebook social graph.

This talk will go into detail on these frameworks and their evolution.