In this seminar, I will present a method that addresses the problem of finding optimal selling strategies for filling a large execution order. If the execution of a large order is not done efficiently, the resulting bad filling prices may have serious consequences on the market equilibrium. I will present an efficient selling strategy based on a trading algorithm that maximises the volume-weighted average price. The method involves Machine Learning tools such as Reinforcement Learning, Q-Learning, and Stochastic Optimal Control techniques. I will present numerical examples showing that the algorithm significantly improves trading performance and regularly meets or even exceeds market benchmarks. Although applied here to stock markets, Reinforcement Learning and Stochastic Optimisation can be used for addressing a variety of problems, including pure learning problems, dynamic resource allocation problems, or sequential decision problem during an epidemic.