Developed a production-ready makeup transfer feature for the lixionga/ProFace repository, focusing on privacy-conscious image processing workflows. Leveraging Python, PyTorch, and deep learning techniques, the work centered on implementing a Pix2Pix-based model combined with a face parsing network to enable realistic makeup transfer between facial images. The solution included a dedicated MakeupPrivacy component to ensure privacy-preserving data handling, along with utilities for image conversion and normalization to support robust performance across diverse inputs. A comprehensive usage example was provided to streamline integration and demonstration, emphasizing end-to-end capability and readiness for deployment without reported bugs during the development period.