Build Stuff'14

Some software problems are difficult; and then, some are insanely hard. In this talk, we will explore one of the latter, the Mona Lisa Travelling Salesman Problem, and how we used modern tools to tackle it. We will demonstrate how we leveraged F# asynchronous and concurrency models in combination with Azure worker roles, service bus and storage, to build a distributed genetic algorithm which can be deployed, scaled and torn down in a couple of clicks - in about 1,000 lines of surprisingly simple code. While the specific problem at hand might be somewhat unusual, the key ideas presented are broadly applicable, and will illustrate how F# and Azure are a deadly combination to build simple applications that scale.