Creating software with ML is becoming increasingly and unnecessarily complex. The amount of tools, frameworks, infrastructure, models and expertise can be overwhelming. This talk will show you how you can start building ML Apps keeping it simple, using DevOps principles that can help you transition and mature your MLOps strategy. We will list a bill of material and showcase an orchestrated pipeline to get data, transform it, allocate GPU infrastructure, train a model while enabling collaboration and deploying an application with ML Backend. This talk will close with a case-study of the Data science team at GitLab where the data science pipeline has been partially automated using DevOps