Worked on the mlflow/mlflow repository to deliver a performance-oriented feature aimed at improving asynchronous trace logging throughput. Focused on backend development and performance optimization, the work involved increasing the default maximum span batch size for async trace logging, guided by internal benchmarks to address logging overhead during high-volume experiment runs. Using Python and leveraging MLflow’s observability framework, the change enabled more scalable trace collection for large-scale workloads. The implementation adhered to established code-quality standards, including proper commit sign-offs and co-authorship. This contribution enhanced the efficiency of trace logging, supporting the repository’s goals for robust and scalable experiment tracking infrastructure.