Sagemaker Jupyter notebook

Some useful points:

• The typical arguments of cloud vs local will apply (as with e.g. Cloud9, Workspaces, etc): Can de-couple your work from the lifetime of your laptop, keep things running when your local machine is shut down, right-size the environment for what workloads you need to do on a given day, etc.
• SageMaker notebooks already run in an explicit IAM context (via assigned execution role) - so you don't need to log in e.g. as you would through the CLI on local machine... Can just run sagemaker.get_execution_role()
• Pre-built environments for a range of use-cases (e.g. generic data science, TensorFlow, PyTorch, MXNet, etc) with libraries already installed, and easy wiping/reset of the environment by stopping & starting the instance - no more "environment soup" on your local laptop.
• Linux-based environments, which typically makes for a shorter path to production code than Mac/Windows.
• If you started using SageMaker Studio, then yes there are some native integrations such as the UIs for experiment tracking and endpoint management/monitoring; easy sharing of notebook snapshots; and whatever else might be announced over the next couple of weeks.