Muvr is a real-time personal trainer system. It must be highly available, resilient and responsive, and so it relies on heavily on Spark, Mesos, Akka, Cassandra, and Kafka-the quintuple also known as the SMACK stack. In this talk, we are going to explore the architecture of the entire muvr system, exploring, in particular, the challenges of ingesting very large volume of data, applying trained models on the data to provide real-time advice to our users, and training & evaluating new models using the collected data. We will specifically discuss how we have used Cassandra to store large volume & velocity of biometric data from consumer-grade and medical-grade sensors, then how to securely perform analysis, clustering and pre-processing these data sets in Spark. We will close by showing how we architected and implemented (R-)CNN models, which are fitted at scale, and then how to persist and deliver the model parameters back to the mobiles for real-time classification. (Demos included!) About the Speakers Jan Machacek CTO, Cake Solutions I focus on large-scale machine learning and big data systems, particularly those that interact with IoT, wearables and mobiles. Naturally, I am a passionate technologist; I have hands-on experience with delivering large-scale systems particularly those that bring together the data science and mathematics with modern engineering practices. I regularly contribute to open source projects, and I speak at conferences in the UK and abroad. Anirvan Chakraborty Technical Director, Cake Solutions Limited Ani, the technical director at Cake Solutions, is a passionate programmer, published author and a big fan of open source software. He takes a keen interest in functional programming and more recently in the DevOps movement. Ani heads up the DevOps team at Cake. Some of his key responsibilities are technical quality assurance, technical resource allocation and recruitment, providing architect support to CTO, strategic consulting, planning and process optimization.