Streamlined Pre-Approved Digital Loans
2019.
Technologies: Python, Pandas, Jupyter, Azure Kubernetes, numpy, Snowflake Data Warehouse
The customer, a financial institution, wanted to offer pre-approved loans through a streamlined digital process. I engineered an ETL pipeline to clean, and join data for 1 million customers. As proof of concept, this system made over 1 million in new loan offers.
I wrote 75 Python scripts for approachable ETL without added infrastructure and processed data from over 20 data sources using Pandas and NumPy. There were difficult splits in the time and order of the data. I extracted valuable features from textual data to develop loan affordability metrics and was able to produce modeling quality datasets for a client-specific loan offer ranking model.
Then I worked to include in internal stakeholders to this new system through database tie-ins to PostgreSQL, Snowflake, Teradata, and legacy systems. Then I helped to design a front-end interface, redefine the customer journey, and advised on pdf contract generation.
The pipeline worked well and was used as the basis to expand and spearhead new digitization efforts.
Do you have an ambitious project? Email me and let us work together: inquire@automatedinnovations.com