This chapter shows how portfolio allocation can benefit from the development of large-scale portfolio optimization algorithms such as the coordinate descent, the alternating direction method of multipliers, the proximal gradient method, and Dykstra's algorithm. With these optimization algorithms, it considers more complex portfolio optimization programs with non-quadratic objective function, regularization with penalty functions and nonlinear constraints. The chapter discusses three main models of smart beta portfolios: the equal risk contribution portfolio, the risk budgeting portfolio, and the most diversified portfolio. A robo-advisor has two main objectives. The first objective is to know the investor better than a traditional asset manager. Because of this better knowledge, the robo-advisor may propose a more appropriate asset allocation. The second objective is to perform the task in a systematic way and to build an automated rebalancing process.