We are interested in optimization problems for which the objective and constraint functions are evaluated by running computer codes seen as blackboxes. In this context where derivative information is not available, blackbox optimization is an option. We consider the Mesh Adaptive Direct Search (MADS) method, a search/poll derivative-free class of algorithms with strong convergence properties. This talk briefly introduces the MADS algorithm and its implementation, the NOMAD software. Real-world optimization problems are presented: The optimal positioning of snow monitoring devices over a large territory, the characterization of objects from radiographs, and the biobjective optimization of aircraft trajectories.