Tom C.Rojo

- Developed a production [ETL pipeline](https://github.com/tommcrojo/student-data-pipeline) using dbt, Airflow, Docker, and DuckDB, consolidating 3 data sources with 100K+ student records and reducing execution time from 15 min to 90s (90% improvement).

- Deployed a PySpark [recommendation system](https://github.com/tommcrojo/admissions-ml-api) on Databricks processing 120K+ student profiles using Medallion architecture, resulting in a 12% increase in contact center conversions.

- Developed ETL connectors using Polars for REST APIs, Google Sheets, and PostgreSQL, achieving a 3.3x performance improvement compared to Pandas.

• Developed regression models to predict delivery punctuality, enabling proactive management of urgent orders and reducing critical delays by 15%

• Consolidated multi-source telemetry data (GPS logs, timestamps, and cargo types) to build a high-fidelity dataset for ML, reducing data inconsistencies by 40% and enabling real-time driver reliability scoring