Hey Nikos,
Actually callbacks are executed on the supervisor host. This is because callbacks are used more for job coordination. However we have had other clients use it to update another database automatically. If you really need qube! to execute a command on a worker, you could chain another job to execute after the first job has finished. I would set it up to run on the host the job was submitted from. However you are correct in your use of the callback code, field. If you take a look at your supervisor's C:\ you'll probably find that file.
Thanks,
Anthony