In January 2026, Khushi integrated HIP backend support into the WASI-NN GGML plugin within the WasmEdge/WasmEdge repository, enabling GPU acceleration for llama.cpp workloads. This work involved backend development and GPU programming, primarily using CMake to ensure seamless integration and maintainability. By unlocking HIP-compatible GPU acceleration, Khushi improved inference throughput and resource utilization for machine learning models running on supported hardware. The implementation adhered to contribution guidelines with a signed-off commit, reflecting attention to code quality and traceability. Although no bugs were reported or fixed, the feature expanded hardware support and laid groundwork for further validation and model coverage.