Naveen Miriyala focused on improving the reliability of benchmarking workflows in the mlcommons/inference repository by addressing a critical issue in MLPerf reference script flag handling. Using Shell Scripting, he ensured that the --accuracy flag was applied exclusively to the accuracy script and removed from the performance script, preventing misinterpretation and incorrect results. He also corrected minor typos in reference_mlperf_perf.sh and reference_mlperf_accuracy.sh, enhancing script stability. This targeted bug fix aligned the scripts with MLPerf standards, improved result integrity, and reduced troubleshooting time for evaluators. Naveen’s work demonstrated careful attention to detail and a strong grasp of Shell scripting practices.