The concept of smart mobility systems is revolutionizing transportation networks worldwide by integrating advanced technologies and innovative transportation services including Mobility-on-Demand and Mobility-as-a-Service. Solving complex decision-making problems in smart mobility systems presents significant modeling challenges arising from intricate human behavior, strategic interactions among multiple stakeholders, and the inherent complexity of transportation networks. Traditional network optimization for designing and operating smart mobility systems often overlooks the significance of endogenous travel demand and the dynamics of multiple agents. However, understanding traveler preferences and guiding their choices is crucial for enhancing system efficiency and moving towards a sustainable transportation framework. This dissertation proposes innovative methods that integrate discrete choice modeling with optimization models, across various decision horizons: operational, tactical, and strategic. First, this dissertation introduces a novel matching algorithm tailored for high-capacity ride-pooling systems. This real-time algorithm, which integrates choice modeling and reinforcement learning, significantly boosts profitability by balancing between immediate acceptance probabilities and the future value of serving requests. Second, a convex program is used to characterize the equilibrium of strategic behavior between a transportation network company and travelers, wherein a company can adaptively adjust pricing and routing strategies while considering travelers’ mode choices. This model can be used as a policy design tool to quantify the impacts of government interventions, including constructing new infrastructure and imposing taxes and regulations. Third, a convex programming formulation is presented characterizing traveler demand patterns by jointly modeling destination, mode, and route choices. This formulation complements the traditional sequential four-step approach to travel demand forecasting by providing robust estimates of choice-related parameters without the need for iterative calibration. Throughout this dissertation, numerical evidence is provided highlighting the benefits of integrating discrete choice modeling with optimization.