Shopify recently developed a new ML platform from OSS components to enable users to efficiently implement and scale a wide range of internal (e.g., fraud detection) and merchant/buyer-facing (e.g., product categorization) use cases. Merlin uses Spark to prepare data and features, which are saved to a data lake or feature store. In addition, each project runs in a dedicated environment, a Merlin Workspace, which is implemented as a Ray cluster deployed to Kubernetes. With Merlin, users create a Merlin Project by specifying what packages and libraries are required for development and Merlin Workspace to define what resources are required. Then, they can easily prototype their ML projects using Jupyter notebooks and move to production by orchestrating and scheduling their workflows with Airflow or Oozie.