During a two-month period, Diego Dupin developed and documented native MariaDB embedding store integration for langchain4j, enabling MariaDB to function as a vector database with support for JSON and column-based metadata storage. He implemented the backend integration in Java and SQL, aligning with existing langchain4j architecture and standards, and included comprehensive tests to validate storage and retrieval workflows. Diego also authored detailed documentation for the langchain-ai/langchain repository, providing installation steps, setup guidance, and practical usage examples. His work improved data locality, enhanced search capabilities, and streamlined onboarding for developers integrating MariaDB vector stores into AI-driven applications.