We study new optimization-driven approaches to two engineering problems, employing techniques from integer, bilevel, stochastic, and multi-objective programming. We first present an approach to the open-pit mining truck dispatching problem that utilizes mixed-integer programming (MIP). The truck dispatching problem seeks to determine how trucks should be routed through the mine as they become available. We describe an optimization-driven approach to solving the dispatching problem in the form of a MIP model. The model is difficult to solve directly, so we present a heuristic that quickly produces high-quality feasible solutions to the model. We give computational results demonstrating the effectiveness of the proposed heuristics and several key model components. To show that our model finds solutions that meet the open-pit mining objectives while accounting for key problem components in novel ways, we embed the MIP-based dispatching policy in a discrete-event simulation of an open-pit mine. We further create two additional heuristic dispatching policies that rely on a new nonlinear rate-setting model that treats queueing at each site as an M/G/1 queue. We present a full computational study of the three policies in which we perform output analysis on key metrics of the open-pit mine simulation. We show that the MIP-based dispatching policy consistently outperforms the heuristic dispatching policies on open-pit mines with a variety of characteristics. The second problem we study is selecting metabolic network changes in cellular organisms. In this problem, enzymes are used to alter the rates at which reactions occur in cellular organisms, causing the cell to increase the output of a desired biochemical product. In existing bilevel MIP models, the lower-level cellular objective is modeled as either maximizing cellular growth or minimizing the biochemical output. We combine these perspectives with two new bilevel MIP models: a single-objective maximum productivity model and a bi-objective maximum yield and maximum growth model. We finally present two-stage stochastic extensions of both models in which we maximize the expected values of productivity, yield, and growth when the planned changes to the metabolic network are uncertain. Because the stochastic bi-objective model contains a complicating budget constraint that lacks parallel structure, we describe a heuristic that alternates between a scenario decomposition-based algorithm and allocating the budget to individual scenarios. Ultimately, we show that this methodology can be implemented to find solutions that meet the metabolic engineering objectives but which are less sensitive to uncertainty than solutions to existing models. This dissertation includes the following three supplemental data files: - A2-SimData.xlsx: data used to generate the simulation of an open-pit mine described in Chapter 3, - A3-CoreModel.xlsx: data used to generate instances of bilevel metabolic engineering models of the core network reconstruction of E. coli as described in Chapter 4, and - A3-iJR904.xlsx: data used to generate instances of bilevel metabolic engineering models of the iJR904 network reconstruction of E. coli as described in Chapter 4.