During July 2026, this developer delivered a nightly benchmark suite for nested vector search in the opensearch-project/opensearch-build repository. The suite evaluates performance across Faiss and Lucene engines using the Cohere 10M dataset, with benchmarks configured to run automatically on r6g.4xlarge instances. Leveraging CI/CD pipelines and Jenkins, the implementation enables continuous monitoring of search performance and rapid detection of regressions. This work enhances visibility into OpenSearch’s nested vector search capabilities and supports informed capacity planning. The focus on performance testing and automation demonstrates a methodical approach to maintaining search quality and operational reliability in production environments.