Haupt Data Analytics & Consulting

Research Associate

LLM Evaluation & Red-teaming

Evaluated LLM accuracy on pharmaceutical data analysis tasks, benchmarking GPT-4 outputs using traditional inferential statistics (t-tests, chi-squared, regression) across 500+ test cases. Used red-teaming and data analytics to assess the performance of AI systems in pharmaceutical data evaluation.

A/B Testing

Conducted A/B testing using Qualtrics for survey administration, and Python and R for data analysis of 10,000+ responses on the University of South Carolina’s ad campaign preferences among panel and list participants.

Natural Language Processing

Extracted 10+ recurring themes from 10,000+ open-ended responses to identify key brand perception drivers for the University of South Carolina. Created large-scale data visualizations using Python and Excel for the study, which was presented to clients.