The study of time series plays a pivotal role in the decision-making process, giving rise to numerous methodologies over time. Within this context, score-driven models emerge as a flexible and interpretable approach. However, due to the significant number of parameters involved, the estimation process for these models tends to be complex. To address this complexity, this study aims to evaluate how the adoption of modern optimization techniques impacts the final performance of the model. Beyond simplifying the parameter estimation process, this shift in paradigm allows for the integration of new techniques, such as robust optimization, into the model formulation, thereby potentially enhancing its performance. The SDUC.jl package, which facilitates the adjustment and prediction of score-driven models based on unobservable components using modern optimization techniques, represents one of the main contributions of this study. By utilizing well-known time series to illustrate its functionality and monthly electrical load data from the Brazilian system, the study was able to demonstrate the flexibility of the package and its robust performance, even during periods of regime change in the data, thanks to the application of robustness techniques.