Anisha Kulkarni enhanced the google-ai-edge/LiteRT repository by updating the OpenVINO compiler plugin to support multiple SoC models, enabling dynamic NPU platform selection during ahead-of-time compilation. She introduced a mapping between soc_model and NPU_PLATFORM using a supported_soc.csv file, integrating this logic into the AOT compile path to ensure correct platform binding based on the selected SoC. The just-in-time compilation flow remained device-specific, preserving existing behavior. Anisha validated the new SOC-based configuration with targeted test harnesses, confirming expected outcomes for different Intel NPU models. Her work leveraged C++, compiler design, and embedded systems expertise to improve hardware portability.