MLOps with Jenkins, MLFlow, Docker, GitHub, and AWS EC2

MLOps is defined as the magical blend of automation, containerization, and occasional frustration.

Introduction

MLOps which stands for Machine Learning Operations is a set of methods for the automation of machine learning processes. For example, if you have a machine learning model on your site you want to keep it updated and retrain once the performance drops. MLOps can help you with this! A possible solution would be setting up a pipeline (a script with workflow instructions) that will automatically monitor the performance of your model and retrain it once it drops, alternatively one can schedule retraining every certain period of time, then the pipeline will try different models, see which one suits better, save it, then evaluate its performance and deploy it to your application without human interaction.

MLOps emerged from the classical CI/CD which is needed for automatic deployment of a usual application. I have already published 3 articles about how to build CI/CD pipelines and one article about MLFlow which automates your machine learning experiments. This article is a logical continuation of them. You can check them out by the links below. And now, ladies and gentelmen, MLOps!