Résumé
Where I've worked, and what I did there. Select any role to open the detail.
- Developed the architecture for and provisioned a secure AI user interface on agentic backend logic to reduce multitenant security risks
- Designed and deployed an e-commerce interface to distribute software and gain business leads for customer acquisition
- Supported a global telecom supply chain organization's ERP system implementation via Boomi integration development
- Produced automated browser-based testing scripts to significantly reduce development time spent on repetitive manual testing
- Designed a cross-platform identity management system integrating Power Apps, Power Automate, Azure Data Factory and SQL Server to centralize Core Identity Directory data from multiple ITSM platforms
- Engineered secure ADF ingestion pipelines using REST APIs and token-based authentication to connect with HRIS services
- Collaborated with a database architect to align the workflow with data warehouse models and standards
- Portrayed agricultural analytics for county and state data for external distribution using visualization techniques in R
- Received honorable mention in the Undergraduate Statistics Project Competition (USPROC) for intermediate statistics
- Collaborated with Evonik to develop construction material compatibility matrices via data scraping and cleaning
- Researched financial machine learning applications alongside faculty Dr. Maxwell Omwenga and Dr. Benjamin Johnson
- Published and presented the research at the 24th Annual IEEE International Conference on Electro Information Technology
- Developed data visualizations of national pest damage and fire damage data using ArcGIS software
Text Extraction and Classification for Automated Balance Sheet Data Entry from Form 1120
May 2024Andrew S. Thompson, Maxwell M. Omwenga, Benjamin W. Johnson. 2024 IEEE International Conference on Electro Information Technology (eIT), pp. 229–234. IEEE.
Data entry is low-skill, high-cost work that stands between accountants and the analysis they are actually paid for. The paper tests automated classifiers for sorting Form 1120 balance sheet items into current and noncurrent assets and liabilities. Logistic regression and linear and polynomial support vector machines proved the most accurate. It is a step toward pulling a whole balance sheet schedule into a spreadsheet unattended.
Read on IEEE Xplore doi:10.1109/eIT60633.2024.10609926
- Directed chapter operations, event planning and financial advising, and fostered an environment of collaborative leadership
- Coordinated fundraising events, surpassing goals and raising over $6,000 for local and national nonprofit partners