Architecting for Machine Learning

Machine learning is presently a key component of software engineering. We have Netflix recommendations, Pinterest image analytics, Duolingo voice synthesis and many other examples demonstrating that machine learning is more than a reality, but already expected by users. The rise of deep learning brings countless exciting possibilities, from video recognition to self driving cars. And everyone can do it! There is a wide variety of open source libraries and public services available to developers. However, the same variety and accelerated pace of inovation can make it difficult to select machine learning technologies and integrate them into applications. Some problems can be solved quite simply by managed cloud services, while others may require training a new model or even developing a custom learning algorithm. Down to the infrastructure level, tuning hardware acceleration with GPUs and FPGAs makes it viable to clean, transform and train unprecedent ammounts of data. This presentation explores the resources available from Amazon Web Services for machine learning, from managed services to libraries and incentive programs. Every developer is welcome, let's get our hands dirty with examples and demonstrations and build ever more intelligent application.