Prasanna Kasar worked on integrating a Keras Model Parser into the SOFIE framework within the root-project/root repository, focusing on production readiness and seamless build integration. Using Python, CMake, and deep learning frameworks, Prasanna implemented support for LayerNorm, BatchNorm ND, and ELU layers, ensuring the parser could handle a broader range of Keras models. The work included creating a dedicated CMakeLists.txt for modular builds and updating Python interfaces to inject the new parser dynamically. By replacing legacy version checks with runtime detection and improving CI traceability, Prasanna delivered a robust, maintainable solution that enhances SOFIE’s machine learning capabilities.