Worked on backend development for the HumanSignal/label-studio repository, focusing on improving the reliability of budgeting calculations. Addressed a critical bug by updating the update_budget_total_spent_from_predictions_meta function to ensure that budget_total_spent defaults to zero when encountering None values. This Python-based solution prevented NoneType errors during cost tracking, enhancing data integrity and reducing runtime failures in production environments. The work centered on bug fixing and backend logic, resulting in more accurate and stable budget reporting. By normalizing None to zero, the update improved the stability of budgeting features and contributed to smoother financial operations within the application.