Timetabling problems are constraint optimization problems proven to be NP-complete. Furthermore, evaluation of violations is costly, and there is no common data format for representing timetabling problem instances. In this paper, a framework for designing memetic algorithms (MAs) to solve timetabling problems is described and a tool, named final exam scheduler (FES) is introduced. FES is the first tool that accepts timetabling markup language (TTML) documents as input. It utilizes an MA with an adaptive violation directed hierarchical hill climbing method for solving examination timetabling problem instances. Experimental results on a set of benchmark data indicate the success of MA.