Comparing a paid vs open-source MLops solution

Our clients are often asking us what is the best technology stack/ architecture to use for their businesses. So today one of our Data Engineers, James Twose, will take us through a comparison of a paid technology stack and an open source technology stack to create, deploy and maintain an MLOps solution. Specifically, a web application that attempts to predict the amount of time the trains in the Netherlands will be disrupted (in minutes) based on the weather conditions on the day (temperature, rain amount).

The following is a presentation of the technology stack comparisons, their learning curves, their pro’s and con’s, and their corresponding prices:

As mentioned in the previous video, here is the setup and demo of the open source, Kubernetes, version of the project:

And finally, the paid, Databricks, version of the MLOps solution:

We hope you found this post enlightening and that what you learned helps you to make decisions for your company, and provided some insights into the process of using these different approaches. 

The GitHub repository for this project, including the presentation, can be found here - https://github.com/cinqict/paid-vs-opensource.