As a Deep Learning Research Engineer at MAPFRE Insurance USA from January 2022 to May 2022, I spearheaded the development of advanced deep learning models to analyze property photos and classify insurance policies based on detected hazards. Utilizing state-of-the-art EfficientNetB5 architecture, I achieved a significant milestone with an accuracy of 66.8% and a recall of 82.4%, while implementing automation capabilities that could potentially reduce manual policy review workload by 78.5%. Additionally, I applied Natural Language Processing (NLP) techniques to cluster over 150 hazards into 5-10 groups, facilitating the development of a prospective multi-class classification model. This innovative approach promises substantial cost savings by streamlining hazard identification and classification processes, with projected savings of $200 per inspection report.