TechTalk Series
When: Wednesday, April 15, 2026
Time: 3:00 pm
Where: RB 104
Title: Towards Automated Data Science Testing: Prototyping a Validation Tool for Silent Bugs in Black Box Scripts
Presented By: Priscilla Zavala – Senior, Ball State University
Abstract: Data integrity often hinges on computational scripts vulnerable to silent bugs (logical errors that compromise results without triggering script failure). This study, mentored by Dr. Ergin, focuses on prototyping an automated validation framework using InterFact railroad monitoring system reports and data. The proposed AI-driven architecture generates synthetic test fixtures based on provided specifications from the developer, injecting targeted errors (such as duplicates or null values) to stress-test black box workflows. This study aims to establish a generalized tool designed to help developers proactively expose hidden logic errors in any analytical pipeline to ensure final insights can be trusted.
Bio: Senior pursuing a B.S. in Computer Science with a concentration in Data Analytics & Machine Learning and a minor in Mathematics. Currently an independent researcher under Dr. Ergin, focused on prototyping an automated, AI-driven validation tool. Outside of academic work, I’m an aspiring Data Scientist who enjoys playing the piano, spending time outdoors, and reading.