In this thesis, we study time-sensitive applications where it is important to minimize the completion time, i.e., time passing between receiving the instance and finishing the implementation of the solution. Different from the traditional approach, we are directly focusing on the minimization of the computation time as well as finding the optimal solution to the problem. The conventional approach to these conflicting objectives is generally to trade off one for the other. As an alternative, we propose a new approach called Computation-Implementation Parallelization (CIP), and develop methods to embed the computation time into the solution-implementation to minimize the total completion time. We implement our CIP approach and show its effectiveness on a type of TSP we call the TSP Race problem, where the goal is to minimize the time between receiving the instance and finishing the travel. We demonstrate a method for determining a priori when CIP will be effective. We also implement our CIP approach on Computation-Time Limited Capacitated Vehicle Routing (CTL-CVRP) problems, and show that it is possible to decrease the computation-only time while maintaining the solution quality. By this means, some of the computation time can be set free and used to improve the customer service either by delaying the order cutoff time or dispatching the trucks earlier. As a tangential study, we develop a new TSP tour length estimation model. Our model is distribution-free, and is shown to produce very accurate estimates on many different node dispersions.