Worked across six repositories, including agiresearch/AIOS, inclusionAI/AWorld, and kubernetes-sigs/kueue, to improve Docker image hygiene and build efficiency. Focused on refining .dockerignore patterns using Dockerfile syntax to exclude Python caches, temporary files, and OS-generated artifacts, which reduced image sizes and improved deployment speed. Addressed misconfigurations in hail-is/hail and standardized exclusion patterns in hardisgroupcom/sfdx-hardis, resulting in more reliable and consistent CI/CD pipelines. Collaborated cross-repo to deliver cleaner builds and fewer cache-related issues, leveraging Docker expertise and pattern-based optimizations. The work enhanced operational stability and deployment velocity while maintaining a focus on maintainable, efficient build processes.