This $506,494 National Science Foundation Project Grant under the Engineering (47.041) program will support research at Purdue University developing novel methods for data-driven engineering of service systems. The research aims to leverage large operational data sets and machine learning technologies to facilitate model identification for complex stochastic network models representing service industries. Methods will be extended for model calibration in random and nonstationary environments. Theory building on areas like M-estimation, minimax analysis, and information geometry will also address identifiability of model classes. The comprehensive "graybox" methodology developed for calibrating service system models is intended to enable engineers, policymakers, and academics to improve operational and cost efficiencies, and consequently national welfare. The five-year award period began May 1, 2022.