Hi

yes, I discovered three working possibilities:

1. Define the pandas functions as inline function as pointed out above
2. Define the pandas function in the same script that is imported as "library" in the dlt config ( 
   
   

   libraries:
   - notebook:
   path: ./pipelines/your_dlt_declaration_containing_high_level_udfs.py​

3. install your python library as whl on the cluster

Regards

DAvid