The increase in complexity of autonomous systems is accompanied by a need of data-driven development and validation strategies. Advances in computer graphics and cloud clusters have opened the way to massive parallel high fidelity simulations to qualitatively address the large number of operational scenarios. However, exploration of all possible scenarios is still prohibitively expensive and outcomes of scenarios are generally unknown apriori. To this end, the authors propose a method based on bayesian optimization to efficiently learn generative models on scenarios that would deliver desired outcomes (e.g. collisions) with high probability. The methodology is integrated in an end-to-end framework, which uses the OpenSCENARIO standard to describe scenarios, and deploys highly configurable digital twins of the scenario participants on a Virtual Test Bed cluster.