Systems, Methods, and Computer-Readable Media for determining attributes of Documents
As a Machine Learning Scientist at Xero, I developed a modified T5 model to extract supplier-names from invoice-type documents. This work was patented which can be seen here: https://patentimages.storage.googleapis.com/66/77/f9/7f76d2a50174b6/WO2025053765A1.pdf
Methods, systems and computer-readable media for training document type prediction models, and use thereof for creating accounting records
During my time as an intern at Xero, I worked on an accounting document type classifier. During my time on this project I ideated and created a ML model architecture and training pipeline that uses multimodal input and an innovative approach to a GRU to create a state-of-the-art solution. This work was patented which can be seen here: https://patentimages.storage.googleapis.com/9e/66/6d/894bb8c032ca37/WO2024043795A1.pdf
