Worked on the mlx-lm repository to enhance type safety in core Python functions, focusing on aligning function signatures with their actual return types. Addressed a key bug related to type annotations, which improved API reliability and reduced the risk of runtime errors. The approach centered on explicit type annotations and refining software development practices to ensure safer, more predictable interfaces. These improvements contributed to greater maintainability and simplified downstream integration for other developers. By clarifying API contracts and improving onboarding, the work provided a more stable foundation for future development, leveraging Python and type annotation expertise throughout the process.