Someone ever tried to 'deploy' a kedro project into Snowpark (as stored procedure?)? With the restriction to the conda snowflake channel in an otherwise sandbox environment, getting kedro itself as well as all the dependencies inside the project itself seems quite a lot to wrap ones head around.

So I think this is probably possible, but it would be a bit fiddly to do so

Full Kedro project - no, but we've been successfully running Kedro nodes (which in fact means offloading the compute to Snowflake clusters) using Snowpark

Perhaps it is indeed a bit too much to ask from the technology, which states to offer 'more general additional capabilities for compute pushdown.' and not end-to-end cloud computing solutions