Deployment of ML models in Azure and Kubernetes

Data Scientists develop a lot of different machine learning models in their projects. Only a few models are chosen for production. Those models need to be deployed as web services to predict new data samples.

We will be providing two approaches to deploy machine learning models to production. One solution uses Azure Machine Learning Service to serve the models. Another solution uses Kubernetes to deploy the models. In both cases we will be using models, which were trained locally and developed with Python.

Furthermore we will show how to define the input, output and the required hardware resources to run the machine learning models as web services in Azure ML Service or Kubernetes.