Wine aging in barrels is a slow, complex process governed by continuous physicochemical transformations, yet traditionally managed through empirical practices and manual inspection. This thesis develops analytical and machine learning models to characterize wine aging dynamics from non-invasive sensing data acquired by the Smart Bung, a patented Internet of Things device by Nova Stark installed on the barrel replacing traditional cork, combined with distributed cellar environmental measurements. A preliminary analysis of cellar microclimatic variability is conducted to establish the spatial boundary conditions of the aging process, informing a representative sensor deployment strategy. Two core modeling components are developed, each motivated by a specific analytical need.