The smart energy sector is poised to deliver huge savings through efficiency improvements and ancillary services such as Open Energi's Dynamic Demand. This talk, aimed at Go programmers who do not have much experience with Machine Learning (ML), will start by introducing common idioms used by well known machine learning frameworks like scikit-learn and Matlab's ML Toolbox, and how they are mirrored in Go packages like Goml and Go Learn. I'll also briefly explain how one can use Storm multilang or os/exec to assemble a polyglot ML solution without requiring a data scientist to make (many) changes to their model code. The second half of the talk will explain how we've leveraged Go primitives and cloud-based services to deliver scalable real-time machine learning services written entirely in Go with low cognitive overhead for both data scientists and developers.