Developed and integrated baseline performance metrics into the continuous integration pipeline for the NVIDIA/bionemo-framework, focusing on the ESM2 model. This work enhanced CI observability by enabling automated validation of execution metrics such as consumed samples, validation loss, reduced training loss, and TFLOPS per second per GPU. Leveraging skills in benchmarking and CI/CD, the developer utilized YAML to configure these metrics directly within the CI workflow. The solution supports data-driven optimization and faster regression detection across model runs, providing a foundation for ongoing performance monitoring and validation within the repository’s automated testing and deployment processes. No bug fixes were recorded.