Worked on stabilizing the Metal backend for Intel Macs across the ggml-org/llama.cpp and Mintplex-Labs/whisper.cpp repositories, focusing on runtime crash prevention and cross-repository consistency. Addressed residency-set initialization issues by implementing conditional compilation, ensuring that residency sets are only created on supported macOS versions and architectures. Utilized C, C++, and Objective-C to deliver concise, well-documented fixes that improved reliability for Intel Mac users. The approach reduced runtime instability and support overhead by preventing crashes related to unsupported configurations. Emphasized code quality and traceability, contributing to a more robust macOS development workflow and consistent user experience across both projects.