Hi @cpatte7372 ,

Serverless compute has some limiations. For instnace, you can't install cluster scoped libraries etc.Take a look at full list here:

- Compute policies
- Compute-scoped init scripts
- Compute-scoped libraries, including custom data sources and Spark extensions. Use notebook-scoped libraries instead.
- Instance pools
- Compute event logs
- Most Apache Spark compute configurations. For a list of supported configurations, see Configure Spark properties for serverless notebooks and jobs.
- Environment variables. Instead, Databricks recommends using widgets to create job and task paramers

Serverless compute limitations - Azure Databricks | Microsoft Learn

But you can install notebook-scoped libraries, like I did below: