Notebook in a databricks cluster keeps a cache while the cluster is running. If you add an import statement and then remove it, the notebook still has a cached instance of that import and will continue to work. Running code in Visual Studio against the notebook creates a new context and therefore the said import is no longer available and blows up with not a useful error. Started building up the notebook on a new one line by line and as soon as I added my culprit line I knew what the issue was.

Before:

After: