dbt + databricks in production: modeling challenges & dbt SQL workflow Question..

I’ve been experimenting with dbt on Databricks while working on a lakehouse-style setup, and I’m curious how others are handling it in real-world projects.

Things I found tricky so far:

• Designing model layers that stay readable over time
• Incremental strategies that don’t turn into a mess
• Balancing performance vs cost on Databricks

One workflow issue I’m also running into:
In VS Code, I can’t reliably run “Execute dbt SQL” to view query output while working on dbt analyses / models. It only seems to work when running SQL from

target/compiled/*

, not directly from the main project files.

How do you usually handle this?

• Do you rely on compiled SQL only?
• Use Databricks SQL editor / notebooks for exploration?
• Or is there a better dbt-native workflow for iterative analysis?

For those using dbt + Databricks in production:
what were your biggest challenges, and what design choices or workflows helped the most?

Would love to learn from the community.