Does anyone have an example of a "foolproof" Snowflake to star job?

By snowflake to star, I mean that you're basically doing some kind of multi-way join on your OTLP database into your OLAP database. So you start with a transformation that's a table input with the fetch, and then you have a table output.

By foolproof I mean the following:

1. Builds the table if its not there.
2. Drops/rebuilds the table if the columns don't match.
3. Fetches the starting point from the destination database (i.e. how far it got last time), then uses that as the starting point for the ETL.

It's 1 & 2 I need the most help with. This seems like a pretty common thing to do.