How would you approach reducing the noise in a continuous stream of data coming from a machine learning model at Omio?
