Worked on enhancing build automation and API stability across the Intel-tensorflow/tensorflow and google-ai-edge/LiteRT repositories. Delivered a Bazel build system upgrade for TensorFlow, updating the tf_keras build script to Bazel 7.4.1 to improve compatibility and support future changes. Focused on future-proofing by refactoring Keras model converter components in both repositories to use public API symbols, reducing reliance on private interfaces and aligning with best practices for maintainability. Leveraged Python, Shell scripting, and CI/CD skills to ensure robust workflows and consistent symbol usage, enabling easier maintenance and reducing risk from upstream changes in Keras and TensorFlow.