Hi Luis, I use dbt for all of my data transformations. I typically use Fivetran or Airbyte to ingest my data.

Because the APIs of the data sources don’t have frequent changes, I haven’t had to worry about schema evolution. However, there have been instances where I have had to create new data connections and resync historical data. You can also choose to update your base models which reference these raw data sources as you see fit.

We use Snowflake for analytical purposes so it wouldn’t make sense to use a data lake. Since data lakes store unstructured data and we don’t use unstructured data in our analytics, it would be more of a pain. A data warehouse adds structure.