Interest in machine learning has exploded recently, and there are many cloud-based solutions that place its power in our hands. However, in applications with limited connectivity, operating on sensitive data, or demanding low latency, it would be preferable to execute such capability on small microcontrollers. Using a simple example project, we examine the Edge Impulse software development kit to see how it supports developers in deploying machine learning capability on tiny, low-power, and economical microcontrollers.