The growing need of responsiveness for manufacturing companies facing market volatility raises a strong demand for flexibility in their organisation. Since the company personnel are increasingly considered as the core of the organisational structures, a strong and forward-looking management of human resources and skills is crucial to performance in many industries. These organisations must develop strategies for the short, medium and long terms, in order to preserve and develop skills. Responding to this importance, this work presents an original model, looking at the line-up of multi-period project, considering the problem of staff allocation with two degrees of flexibility. The first results from the annualising of working time, and relies on policies of changing schedules, individually as well as collectively. The second degree of flexibility is the versatility of the operators, which induces a dynamic view of their skills and the need to predict changes in individual performance as a result of successive assignments. We are firmly in a context where the expected durations of activities are no longer predefined, but result from the performance of the operators selected for their execution. We present a mathematical model of this problem, which is solved by a genetic algorithm. An illustrative example is presented and analysed, and, the robustness of the solving approach is investigated using a sample of 400 projects with different characteristics.