Shahar Mendelson's research has led to ground-breaking progress in the understanding of theoretical data science. He discovered that key questions in data science are actually natural problems in high-dimensional geometry. He then used this geometric interpretation to solve fundamental questions in theoretical machine learning, sparse recovery in signal processing and high dimensional statistics. The machinery he developed has also resulted in the solution of several significant open problems in pure mathematics, mainly in asymptotic geometric analysis and the asymptotic theory of random matrices.