scikit-prometheus
We love scikit-learn as it makes it really easy to build interesting and useful ML models. The goal of this project is to make it easy to add monitoring to your existing scikit-learn pipelines. We do this by extending existing scikit-learn components and and build new ones that expose Prometheus metrics that keep track of, for example the model predict latency, histograms of predict probabilities and many others.