In the government agencies, civil servants are required to have competence or ability to finish the work effectively and efficiently. In fact, the decision-making system for determining position and assignment of civil servants' functional works is still performed manually, so it takes a longer time. Moreover, the results are not totally accurate in terms of their competency. Rough set, hereinafter called Single Rough Set, is a common method to solve this problem, but the process may be very complex and still has the unclassified result. In this research, Multi Rough Set and Majority Vote technique are proposed to enhance system performance of single rough set with multi attributes of job competency. It obtains accuracy rate with 5-fold cross-validation that is 83.67% better than a Single Rough Set and it has 0.947 Area Under Curve (AUC) derived from Receiver Operator Characteristic (ROC). Thus, it can be said that the system performance of Multi Rough Set can be considered excellent in classifying job competency for civil servants' functional works.