Aaron Higuera developed and delivered a model inference configuration for atmospheric neutrino processing in the DUNE/dunereco repository. He updated the CVN LBL model’s inputs and outputs, integrating a new TensorFlow protobuf-based runtime to ensure compatibility with evolving data processing requirements. By disabling bundle usage, Aaron aligned the inference path with the latest protobuf runtime, enabling correct event processing and preparing the system for production pipelines. His work focused on configuration management and machine learning model deployment, utilizing FCL and TensorFlow technologies. The changes addressed evolving runtime needs and improved the robustness of atmospheric neutrino event processing without introducing new bugs.