Worked on the google/langextract repository to enhance the configurability of OpenAI integration for enterprise deployments. Developed a feature in Python that introduced an optional model_url parameter to the OpenAILanguageModel class, enabling users to specify custom OpenAI API endpoints. The implementation focused on robust API development, ensuring that the client initialization dynamically included the base URL when model_url was provided. Comprehensive unit testing and mocking were used to verify correct behavior across different configurations, supporting business needs for customizable AI interactions. The work emphasized maintainable code and strong test coverage, addressing flexibility requirements without introducing regressions or breaking existing functionality.