Worked on enhancing data retrieval integrity and vector reference capabilities within the Weaviate ecosystem, focusing on both the weaviate/weaviate and weaviate/weaviate-python-client repositories. Addressed a metadata extraction issue by ensuring the correct linked class context was used for property retrieval, improving backend reliability. Introduced support for selective vector retrieval in reference queries, allowing users to specify named vectors when fetching references. Leveraged Python and Go for backend development, API integration, and comprehensive integration testing. Refactored test infrastructure to increase cross-version coverage and reliability, reducing regression risk and supporting robust, vector-enabled features across diverse environments and client versions.