Joaquin Parra

introduction

Welcome! I am an undergraduate student at The University of Chicago, pursuing a B.S. in Computational and Applied Mathematics (expected June 2028). I am an Odyssey and Rodriguez Pastor Scholar, and a Dean's List awardee.

ongoing project

This summer, I am working on comparing mitigation methodologies for emergent novel biases in LLM decision-making, advised by Professor Xuechunzi Bai. Recent work suggests that LLMs may not just reproduce inherited social biases, but also generate new ones through adaptive exploration in sequential decision-making. In this sense, the study compares strategies — exploration-based prompting, adversarial debiasing, and reward modeling — to evaluate which interventions can best redirect these dynamics before early patterns become increasingly stable stereotypes.

interests

My broader interests lie in machine learning techniques and their social implications. I am happy to start learning and experimenting deeper with NLP and its implications in human-computer interaction in social terms, as well as AI safety and security — I participated in UChicago XLab's course on the Fundamentals of AI Safety.

Outside of research: soccer, pole vaulting, classical piano, jazz (bossa nova), and Japanese city pop.

past projects

(WIP)