Navigating the intricacies of modern security problems often involves grappling with stochastic, uncertain, and non-stationary elements, compounded by the presence of competing agents who have disparate knowledge and varied degrees of rationality. Despite these complexities, the persistent demand for logical and actionable decision support from senior leaders underscores the necessity for bespoke solutions, frequently requiring tools native to alternative disciplines. Addressing these imperative calls for an eclectic methodological perspective, drawing from diverse disciplines such as operations research, statistics, behavioral economics, and computer science, along with alternative schools of thought within each. To illustrate this approach, we examine a subset of contemporary problems with significant real-world implications. Beginning with our recent work in adversarial machine learning, we emphasize a shift from the prevalent algorithmic culture to a data-modeling culture, contrasting with much of the existent literature. We employ various canonical operations-research methods to attack a decision-maker’s inference on multivariate-Gaussian, hidden-Markov, and generic-Bayesian models. Shifting focus to the contemporary USAF pilot shortage, we discuss initial steps for its alleviation through the combined use of alternative data-science techniques. Thereafter, we delve into two quantitative methods for planning security operations, capitalizing on Bayesian representations of uncertainty as well as behavioral- economic models that capture an opponent’s boundedly rational actions.