Well, this might or might not work for you depending on the particulars of your environment and requirements...but we had a similar thing we needed to do (if i am understanding what you are saying) and here is what we did.
First, we created an active directory group that had permissions that were needed to all the various directories. Then we added a specific user to that group. We setup a scheduled job that ran an executable every x seconds. The scheduled job was setup to run with the id we had added to the active directory group so that the executable would inherit the permissions it needed to the various directories.
The executable reads "job" files that are placed in a directory. The job file tells the executable what file we want to play with and what operation we want it to do, etc...
The web interface is used to gather the requirements from the user and then write a job file.
As you can see, there is a small delay between when the user submits the information and when the actual work takes place, so this might not be right for you if you need a response back to the user from the web app. In that case, you might consider a web service installed on your remote server. You could then send requests from your web interface to the web service and your web service could then manipulate the file and return a success\failure message or something similar.