Developed a model validation enhancement for the microsoft/onnxruntime-genai repository by implementing perplexity metrics within the existing validation workflow. This work focused on integrating automatic perplexity calculation to support comprehensive performance reporting and improved model monitoring, addressing governance requirements for model risk assessment. Leveraging Python for scripting and data analysis, the solution was validated within the current pipeline to ensure seamless adoption and reliable evaluation of model quality. The approach demonstrated proficiency in machine learning and model validation, enabling more robust release validation and monitoring processes while contributing to the overall supportability and transparency of model evaluation in production environments.