abstract: Cells specific geometries and organizations are most of the time the main actors shaping their function. In the same time, taking into account such complex geometries in stochastic or PDE models usually leads to heavy simulations and limited analysis. In the context of vesicular release at neuronal synapses, we used conformal mapping and asymptotic analysis to compute the narrow escape time at a cusp-like geometry. We used this result to build a simple discrete-continuum model that takes into account both a stochastic regime governed by rare events and a continuous regime in the bulk. With this model, we then investigated the mechanisms inducing paired-pulse facilitation and depression at neuronal synapses.