Worked on the elastic/elasticsearch repository to deliver GP LLM v2 inference support, introducing the gp-llm-v2 model ID and a pre-configured inference endpoint to the backend inference service. The implementation involved refactoring variables and settings for improved clarity and maintainability, as well as updating mocks and mock responses to ensure compatibility with the new model. Leveraging Java for backend development and API integration, the work aligned completion service configurations to support seamless rollout and backward compatibility. Emphasis was placed on maintainable code through consistent naming and autoformatting, enabling rapid experimentation and preparing the system for customer-ready deployment scenarios.