Kiera McCormick

Kiera McCormick

Computer Science MS Student | Human Language Technology

Johns Hopkins University

About Me

I am a Master's student in Computer Science at Johns Hopkins University, specializing in Human Language Technology. My research focuses on developing and evaluating Large Language Models for scientific research, particularly in astrophysics.

I'm passionate about bridging the gap between AI capabilities and domain expert needs, working at the intersection of natural language processing, machine learning interpretability, and scientific applications. I'm also interested in AI safety and the social ethics of technology.

Research Interests

Large Language Models • AI Safety and Misalignment • Scientific AI Applications • ML Interpretability • Human-AI Collaboration • NLP Evaluation • Prompt Engineering

Experience

Research Assistant

AI for Forensics (AI4F) – JHU / NIST

January 2026 – Present

Conducting research at the intersection of AI and forensic science in collaboration with NIST. Work includes surveying AI tools across forensics-adjacent domains, building structured analyses of AI adoption patterns, and contributing to stakeholder documentation on responsible AI integration in forensic laboratories.

Research Intern

Center for Astrophysics | Harvard & Smithsonian - AstroAI

Summer 2025

Conducted research on Multi-Modal LLMs for Astrophysics, focusing on prompt engineering to generate meaningful summaries about astrophysical sources, extracting embedded physical parameters, and applying sparse autoencoders to uncover patterns and relationships between astronomical objects.

Research Participant

Jelinek Summer Workshop & Space Astronomy Summer Program

Summer 2024

Developed and evaluated LLM-powered chatbots for astronomy research, deployed in Slack. Created evaluation datasets with expert astronomers and implemented LLM-based relevance scoring systems to assess chatbot response quality.

Data Science Intern

Los Alamos National Laboratory - Omega Technical Services

Summer 2023

Collaborated with the industrial engineering team on task scheduling optimization through databases and automation. Utilized MicroStrategy, Excel, SQL, and Python for data analysis and visualization.

Publications

Encoding and Understanding Astrophysical Information in Large Language Model-Generated Summaries

McCormick, K., Martínez-Galarza, J. R.

NeurIPS: ML4PS Workshop, 2025 (accepted)

From Queries to Criteria: Understanding How Astronomers Evaluate LLMs

McCormick, K., Hyk, A., Zhong, M., Ciucă, I., et al.

Conference on Language Modeling (COLM), 2025 (accepted)

Designing an Evaluation Framework for Large Language Models in Astronomy Research

Wu, J. F., Hyk, A., McCormick, K., Ye, C., et al.

arXiv, 2024

pathfinder: A Semantic Framework for Literature Review and Knowledge Discovery in Astronomy

Iyer, K. G., Yunus, M., O'Neill, C., Ye, C., Hyk, A., McCormick, K., et al.

ApJS 275 38, 2024

Selected Projects

Testing Social Reasoning in LLMs

Spring 2026

Designed and evaluated a multi-dimensional benchmark of 40 original scenarios spanning Theory of Mind, Pragmatics, and Chain-of-Thought reasoning across 12 state-of-the-art LLMs. Recruited a human baseline (n=12) to quantify the gap between model and human-level social reasoning.

Regional Variation in ChatGPT Use Across U.S. States

Spring 2026

Applied BERTopic to 68,668 first-turn prompts from the WildChat corpus to study geographic and socioeconomic variation in ChatGPT use across 36 U.S. states. Found that education and income correlate with specialized creative uses, while generic image-generation requests are more common in lower-education and lower-income states.

A2CPS Discovery Agent

Spring 2026

Built an AI-powered natural language interface for querying the A2CPS biomedical dataset, translating plain English questions into read-only SQL and surfacing relevant documentation using retrieval-augmented generation. Developed with React, Hono, and the Anthropic Claude API.

Hotspot Homing Robot

Sept 2024 – May 2025

Autonomous mobile robot for detecting and localizing rogue access points in office environments. Integrated wireless signal processing with SLAM navigation systems. Sponsored by RTX.

Solar Panel Mars Rover

Fall 2023

Autonomous Mars Rover prototype with solar recharging capabilities. Developed motion control system using Arduino and implemented sun-tracking algorithm for optimal energy collection.

Quantum Computer Simulator

Spring 2022

Python-based Quantum Computer simulator leveraging object-oriented programming principles in a collaborative semester-long project.

Get in Touch

Feel free to reach out! I'm always interested in discussing research collaborations or interesting problems at the intersection of AI and science.

📧
Email

kmccor23 [at] jh [dot] edu
kieramccormick6 [at] gmail [dot] com