Hey @adrianhernandez ,
Technically yes, but itโs not recommended.
You could technically build everything needed to compile the wheel directly from a Databricks notebook using a setup.py, and store it in a volume, CodeArtifact, or any supported cloud storage location โ provided you have the right permissions (for example, via an instance profile and a dedicated or single-user cluster).
 
However, this approach will likely cause issues in the long run. You wonโt be able to properly version or validate your builds, youโll lose traceability of changes, and eventually, someone will overwrite a working wheel with a broken one โ which will be painful to troubleshoot.
 
My recommendation is to handle it through a proper repository with version control and a CI/CD pipeline that validates the wheel, tags the release, and publishes it to an artifact repository (e.g., CodeArtifact, Artifactory, or Nexus) that acts as a proxy for all users who need to install it.
 
This way you maintain build consistency, version traceability, and deployment reliability โ without the hidden risks of manual or in-notebook builds.
Hope this helps :),
Isi