Nikolai Pak

•Lead a team of 2 people that was responsible for developing and supporting models for document data extraction and collaborated with other teams •Led the development of a new solution based on combining LLM and Image based neural network •Identified common issues and implemented solutions to enhance accuracy of both table and non-table data extraction models, and expanded their application in new scenarios •Refactored and rewrote a substantial portion of legacy code to enhance its maintainability

• Developed and supported ML models for table extraction, document data extraction and classification Improved quality and extended functionality of existing solutions • Implemented multimodal document classifier to boost classification accuracy • Created prototype for non-table data extraction that was used as the foundation for a new solution