API Strategy & Practice Conference 2017

"The world is full of seas of data. Some of these seas are created and exchanged by APIs; some of them are even about APIs. Since 2014, APImetrics has accumulated over 100 GB of data on API test calls made to over 5000 API endpoints by agents deployed in cloud locations on 5 continents.

There's a huge amount of insight trapped at the bottom of that sea of (in our case unlabelled) data. Getting at it would've been nearly impossible before the emergence of powerful open source deep learning libraries in the mid-2010s.

APImetrics will share how we chose a deep learning library and the data munging we did to get our data to work with the library. We will explain how we were able carry out unsupervised and semi-supervised learning and discuss the insights on global API performance and quality we were able to dredge from the bottom of our sea of data. We'll provide pointers on how organizations, from startups like APImetrics to megacorporations, can use deep learning to create oceans of knowledge from their own seas of data."