This question has several flavors. I’m following the AWS Glue flavor right now. Since Glue does not have interactive notebooks (as far as I know), you would have to set up another DataContext on your dev machine and configure the expectation store to point to the same location on S3 that the AWS Glue job is pushing to. Then you should be able to run a Jupyter notebook on the dev machine for editing the expectations. This is probably not an ideal solution. We’re working on a how-to-guide for a nicer workflow with AWS Glue that should help.