This is a simple example workflow for the deployment of a multivariant time series, LSTM based, recurrent neural network. It is based on the bike demand predition dataset from Kaggle and uses the trained model to predict the demand in the next hour based on the demand and the other features in the last 10 hours.