DevConf.US 2021

Have you ever experienced a drop in model accuracy or performance when moving your model from a data scientist's hand to AI DevOps engineer's hand putting it in production? What if data scientists and AIDevOps engineers can receive feedback faster, already when the data scientist changed something in the model or the dataset used to create that model or some packages in their software stack have been changed by pipelines and bots? What If we want to know what is the best place to deploy an application. In this talk, you will discover how automation through Tekton pipelines for MLOps can boost the productivity of the team working on ML projects by deploying the best model satisfying business and technical requirements.