Google's Tensor Processing Units (TPUs) are revolutionizing the way data scientists work. Week-long training times are a thing of the past and many models can now be trained in minutes in a notebook. Agility and fast iterations are bringing neural networks into regular software development cycles and many developers are ramping up on machine learning. This session will introduce TPUs, Edge TPUs then dive deep into their microarchitecture secrets. It will also show you how to use them in your day-to-day projects to iterate faster. In fact, we will not just demo but train most of the models presented in this session on stage in real time, on TPUs & Edge TPUs.