I work with methods for uncertainty quantification (UQ), such as the multilevel delayed acceptance extension of the Metropolis-Hastings algorithm, and apply them to complex numerical models used in different fields of research, such as astrophysics. Such models are often too expensive for traditional UQ techniques, which is why we plan to develop and apply methods to make further use of parallelization where possible. My goal is to answer theoretical questions and explore new applications of UQ methods.