Rare event evaluation is ubiquitious in modern system design. With growing complexity at the nano age, aleatory and epistemic uncertainty render system stochastic and its rare event evaluation, such as failure probability, increasingly prohibitive. To address such challenge, academia and industry have developed on various sampling technique and surrogate models. In this talk, we will present three recent hybrid frameworks to efficiently estimate and optimize the probability of rare events, which is often associated with reliability analysis and circuit yield in Electronic Design Automation.