Worked on the move-coop/parsons repository to address a critical issue in BigQuery data handling, focusing on type coercion for columns containing both integers and floats. The solution involved updating the logic to default to float types in mixed numeric columns, thereby preserving numeric precision and preventing data truncation that could impact downstream analytics. This fix improved data reliability for users integrating with BigQuery. The work was implemented using Python and leveraged skills in data type coercion and database integration. The approach demonstrated careful attention to data integrity and ensured that analytics pipelines maintained accuracy when processing heterogeneous numeric data.