`pc.remote(my_udf, schema)`
Where
`def my_udf() -> DataFrame`
We link the appropiate Python version at cluster startup.