During May 2025, work focused on enhancing the liguodongiot/transformers repository by upgrading the CI Docker image to improve compatibility and performance for AMD hardware. This involved updating the Docker base image to a newer PyTorch version, leveraging containerization and CI/CD practices to streamline continuous integration workflows. The technical approach centered on optimizing AMD CI testing, ensuring more reliable and faster feedback during development cycles. By addressing hardware compatibility and base-image management, the changes reduced CI time and expanded test coverage across AMD platforms. The work demonstrated proficiency in Dockerfile management, DevOps methodologies, and the integration of modern CI/CD pipelines.