This document introduces Auto ML and compares traditional machine learning approaches to Edge ML. Traditional ML has limitations including not scaling well with large data as the number of parameters increases. It also does not provide a way to choose the best family of models without extensive grid search. Regular Auto ML approaches try to address these problems but still require large hardware resources. Edge ML takes a Bayesian approach called MODL that avoids grid search and scales to tiny hardware resources while producing accurate and robust models.