The resource-constrained project scheduling problem (RCPSP) is a well-known problem in project management which aims to determine a schedule and an assignment of activities on resources over a schedule horizon. It is a particular case of scheduling problems where: (i) resource capacities and availabilities are limited, (ii) precedence links applied between the activities and, (iii) the objective is, generally, to minimize the completion time, define as the earliest time in which all the activities are completed. In this paper I study a RCPSP with industrial constraints including multi-skill, multi-project and time windows. It is solved using a genetic algorithm and numerical experiments are conducted on instances of different size including industrial data sets.