When analyzing big data, you often have to process data at scale that's not rectangular in nature. You'd like to scale out your existing programs and cognitive algorithms to analyze that data. To address this need and make it easy to add domain specific code, U-SQL includes a rich extensibility model that lets you process any kind of data, from CSV files over JSON and XML to image files. You can even add your own custom operators. This session will provide some examples on how to use U-SQL to process interesting data formats with custom extractors and functions, including JSON, images, use U-SQL's cognitive library and finally show how U-SQL lets you invoke custom code written in Python and R.