Sequential design methods have received a lot of attention in the computer experiments community, to address designs problems where the task is to estimate some "local" features of an expensive-to-evaluate computer model. Typical examples of such features include : maximizers and/or maxima of the function (optimization), failure probabilities or quantiles (reliability analysis), and so on and so forth. A popular approach to construct such sequential designs is to adopt a Bayesian point of view : the computer model is endowed with a prior distribution (typically, a Gaussian process prior) and a "sampling criterion" (aka acquisition function, or infill criterion) is defined, using the posterior distribution, to quantify the expected benefit of a new evaluation at a given location. This talk will provide an introduction to these methods, and then discuss how Sequential Monte Carlo (SMC) methods can be used to implement them efficiently. The Bayesian subset simulation (BSS) algorithm will be presented as an example.