Consider another approach, using a Configuration Management tool. One of which might suit your needs and is more powerful: Puppet.
Let the guy put all configuration to be done in a Puppet file. Advantages:
- A very robust way of managing the configuration changes. All is in a file, rather than describing what has been done.
- You will stay in control, as it lets you...
- try the configuration on another machine and verify it.
- decide to apply it whether it looks good or not.
- pick the moment in time the changes take place
- Reduce the time the changes are happening. It's automated.
- You can repeat a similar configuration very easily just by customizing the configuration definition.
It boils down to this: the person responsible for the actual application configuration does not need shell access and it provides a way of separating responsibilities of him and the system administrator(s).
Big practical disadvantage too: it's a big elephant to smash this mosquito, so it might not fit your current situation. Not everyone is familiar with Puppet and low-quality Puppet definitions are worse than having a script setting it all up.
This leads me to another similar but simpler approach: let the guy develop a script that runs the necessary commands to change the configuration. It has some of the advantages listed above, yet does not require that much of an elephant of a tool.