This thesis presents the development of a software interface designed to streamline the use of Turbospectrum, a tool for generating synthetic stellar spectra. The primary objective was to create a user-friendly and adaptable program capable of automating the generation of large volumes of synthetic spectral data with Turbospectrum. The data can then be used for training artificial neural networks in spectroscopic analysis of M-dwarfs, as part of the Department of Physics and Astronomy's project within the AI4Research initiative at Uppsala University. The software was developed in Python, chosen for its simplicity, extensive libraries, and familiarity to the user base. The architecture was designed to be modular, facilitating easy maintenance and future extensions. The main modules include functionality for configuration setup, parameter generation, Turbospectrum integration, and output management. Each module was rigorously tested through unit and integration tests to ensure proper functionality and interaction. To validate the program, both functional and non-functional requirements were assessed. Feedback from regular meetings with the project supervisor and usability testing with potential users ensured that the software met the user needs and was easy to use. Performance tests demonstrated that, despite not being the primary focus, the developed software maintained acceptable performance levels. Overall, this project successfully created a flexible and efficient interface for Turbospectrum, enhancing its usability for large-scale spectroscopic studies and advancing AI-driven research in astronomy