Worked on targeted improvements in build optimization and image processing reliability across red-hat-data-services/vllm and liguodongiot/transformers. In vllm, optimized the Docker image by excluding system documentation during build, reducing image size and deployment time. In the transformers repository, addressed robustness in the mllama image processing module by refining handling of image dimensions, impractical aspect ratios, and default PIL formats. Enhanced reliability through expanded unit testing, validating changes across diverse scenarios. Leveraged Dockerfile and Python expertise to streamline build artifacts and minimize runtime failures, resulting in faster deployments and more dependable image processing workflows for both repositories within the month.