Each project in AI Workbench has its own isolated containerized environment where you can run your code and apps. This means that you can work on your project without worrying about affecting other projects or environments. It is also what enables the portable nature of AI Projects and the ability to move between compute resources.
To customize your project’s environment, you can use familiar files (e.g. requirements.txt) and interfaces. For example, you can install packages, write scripts, configure environment variables, add bind and volume mounts, and more. This allows you to tailor your project to fit your specific needs.
Once you’ve made changes to your project’s environment, AI Workbench will automatically generate a Dockerfile/Containerfile for you and use it to build an image for your project. This image will include all the changes you’ve made, so you can easily test and share your configuration.
Some changes you make to your project’s environment may require that your container is rebuilt to see them take effect. For example, if edit an installation script, you’ll need to rebuild your container to pick up the changes. However, not all changes require a rebuild - for example, if you simply add a new environment variable, you just need to restart the container, not rebuild.