Worked on the opensearch-project/k-NN repository to deliver a memory-optimized search encoding feature that supports both integer and long sequences while mitigating overflow risks. The approach involved implementing a new encoding method in Java, enabling the indexing and querying of larger datasets without encountering overflow errors. Comprehensive unit tests were developed to validate the correctness of the new encoding across various sequence types, ensuring reliability for production workloads. The work also included documentation updates and integration sign-off to reinforce code quality and traceability. This contribution focused on back end development and enhanced the scalability and stability of the k-NN module.