In January 2025, Behdad contributed to the Esri/harfbuzz repository by implementing targeted Apple SIMD optimizations for the hvgl table, focusing on performance improvements for little-endian Apple hardware. Using C++ and low-level programming techniques, Behdad removed hard-coded gating for Apple SIMD and introduced robust byte-order checks, ensuring that SIMD instructions are only applied on safe architectures. This approach reduced the risk of incorrect optimizations across devices and improved build system flexibility. The work demonstrated a strong understanding of performance optimization and architecture-aware programming, resulting in safer, more efficient code paths for the hvgl table without introducing new bugs or regressions.