In December 2024, Abel Linares enhanced the rapidsai/rmm repository by implementing fabric handle and memory protection flag support for the cuda_async_memory_resource component. He updated the allocation_handle_type enum to include fabric types and introduced a new access_flags enum, enabling precise control over GPU memory accessibility. Working primarily in C++ and CUDA, Abel focused on low-level programming and memory management to improve both safety and configurability of GPU allocations. He also developed environment-aware tests that skip unsupported configurations, ensuring reliable continuous integration. The work demonstrated depth in system-level engineering while maintaining API compatibility and addressing the evolving needs of GPU resource management.