Antonio worked on the Unstructured-IO/unstructured repository, focusing on improving ontology image categorization within HTML structures. He addressed a bug where images inside div or span elements lacking text were misclassified, leading to inaccurate ontology annotations. Using Python and leveraging his expertise in HTML parsing and ontology mapping, Antonio implemented logic to ensure such images are correctly identified and annotated as images. He also developed targeted tests to cover scenarios involving empty-text containers, safeguarding against regression. This work enhanced the accuracy of downstream data extraction and ontology alignment, contributing to higher data quality in document processing workflows for HTML-derived content.