OCR & Document Extraction using vision models