Worked on the thingsboard/langchain4j repository to enhance Azure OpenAI integration by enforcing explicit temperature parameters across chat, language, and streaming models. The approach involved removing default temperature values, requiring users to specify this parameter directly, which improved model predictability and consistency for both synchronous and streaming interfaces. Focused on backend development and API integration using Java, the changes standardized parameter handling and error reporting across modules. This update enabled more reliable enterprise deployments and simplified testing and quality assurance processes. The work addressed cross-module consistency, ensuring that model behavior remained uniform and transparent throughout the Azure OpenAI integration.