During February 2025, work centered on enhancing the ggml-org/llama.cpp repository by implementing quantization for the CLIP Visual Projector, targeting improved model efficiency and reduced deployment size. The approach involved developing a dedicated command-line interface tool in C++ to streamline the quantization process, making it more accessible for users working with CLIP-based visual projects. Documentation was updated to provide clear guidance and best practices for utilizing the new quantization feature. This contribution leveraged skills in C++ development, CLI tool creation, and model optimization, resulting in a reproducible workflow that supports efficient inference and deployment of machine learning models.