We study the problem of allocating projects to heterogenous workers. The simultaneous execution of multiple projects imposes constraints across project teams. Each worker has preferences over the combinations of projects in which he can potentially participate and his team members in any of these projects. We propose a revelation mechanism that is Pareto-efficient and group strategy-proof (Theorem 1). We also identify two preference domains on which the mechanism is strongly group strategy-proof (Theorem 2). Our results subsume results by Monte and Tumennasan (2013) and Kamiyama (2013).