This talk presents a prescriptive analytics framework for designing inspection and response operations of service utilities facing risks of natural disasters and security attacks. In the first part of the talk, we introduce a stochastic orienteering and network probing problem for localizing failures in the aftermath of a major natural disaster. We develop a predictive model for failure localization using data from inspection operations of a real-world natural gas pipeline network. Next, we exploit the problem structure to design a scalable non-adaptive algorithm based on integer programming. Our results lead to practical and efficient strategies for disaster response operations. In the second part of the talk, we consider the problem of monitoring vulnerable network components with the minimum number of smart detectors against multiple random or adversarial disruptions. This network inspection problem can be formulated as a mathematical program with constraints involving Nash equilibria of a large-scale strategic game. We develop a scalable approach that computes randomized monitoring strategies based on solutions of a minimum set cover problem and a maximum set packing problem, along with optimality guarantees. We demonstrate that the proposed framework effectively utilizes the available network data  to improve the resilience of critical infrastructures against a broad class of failures. The first part is joint work with Mathieu Dahan and Georgia Perakis, and the second part is joint work with Mathieu Dahan and Lina Sela.