Worked on the IndicoDataSolutions/indico-client-python repository to enhance OCR-driven data workflows, focusing on improving dataset creation and training reliability. Expanded the range of OCR engine options, allowing for more flexible and configurable dataset preparation pipelines. Addressed issues in model training options and submission-wait logic, which reduced end-to-end latency and improved the robustness of the training and submission process. Leveraged Python for backend development, API integration, and data processing, ensuring that new OCR tooling aligned with evolving workflow requirements. The work delivered a more reliable and adaptable foundation for OCR-based data preparation, supporting advanced automation and streamlined data handling.