Worked on the microsoft/presidio repository to deliver a flexible Docker configuration system for NLP models, focusing on improving deployment reliability and reproducibility. The approach prioritized inline configuration settings over external files, reducing configuration drift and ensuring that Stanza models remained accessible within containerized environments. Using Python and Docker, the solution streamlined the onboarding process for new NLP models and accelerated experimentation by aligning configuration resolution with inline settings. This work addressed configuration file handling and model accessibility, enhancing the reliability of the NLP engine in production and simplifying the management of complex deployments through improved configuration management practices.