In addition to the other suggestions:

Review the code for places where sentences are being constructed in
code, such as "Welcome " + username +", what do you want to do?".
Replace them with a string format operation that will allow the
localizer to move the sentence parts around.

If your application uses a database, review how one assigns a different
collating sequence, and decide how you'll handle this in deployment.

If your target locales will (or may) include ones that require
multibyte characters, review your code for how you'll implement this.
My personal recommendation is full Unicode support, if it is possible.

Investigate the use of Translation tools. Some of them will provide a
pseudo-translation feature that will allow you to easily check your
process in the development stages.