This thesis is dedicated to the study of stochastic processes; non-deterministic physical phenomena that can be well described by classical physics. The stochastic processes we are interested in are akin to Brownian Motion and can be described by an overdamped Langevin equation comprised of a deterministic drift term and a random noise term. In Part I we examine stochastic processes in the Mesoscale. For us this means that the Langevin equation is driven by thermal noise, with amplitude proportional to the temperature and there exists a genuine equilibrium thermal state. We apply a technique known as the Functional Renormalisation Group (FRG) which allows us to coarse-grain in temporal scales. We describe how to obtain effective equations of motion for the 1- and 2-point functions of a particle evolving in highly non-trivial potentials and verify their accuracy by comparison to direct numerical simulations. In this way we outline a novel procedure for describing the behaviour of stochastic processes without having to resort to time consuming numerical simulations. In Part II we turn to the Early Universe and in particular examine stochastic processes occurring during a period of accelerated expansion known as inflation. This inflationary period is driven by a scalar field called the inflaton which also obeys a Langevin equation in the Stochastic Inflation formalism. We use this to study the formation of Primordial Black Holes during a period of Ultra Slow-Roll. We finish this thesis by applying the techniques developed in Part I to a spectator field during inflation. FRG techniques can compute cosmologically relevant observables such as the power spectrum and spectral tilt. We also extend the FRG formalism to solve first-passage time problems and verify that it gives the correct prediction for the average time taken for a field (or particle) to overcome a barrier in the potential.