The analysis of massive amounts of data sent in real-time by sensors has experienced a major development in the last few years. Due to data heterogeneity, the application of ma- chine learning models specifically calibrated for accurate use cases allowed to extract and infer valuable information. However, few systems propose a distributed implementation on a true industrial cluster permitting of taking advantage of increased computing capabilities. Here we present a demonstration of anomaly detection on an underground drinkable water network located in île-de-France, realized with an innovative platform: WAVES.