Work Experience

Research Assistant
University of Notre Dame
· Notre Dame, IN
Aug. 2021–Present
  • Advised by Dr. Jane Cleland Huang, conducting research to create safer software systems through improved software engineering practices.
  • Applied NLP techniques to improve trace link prediction and support automated software documentation.
  • Developed a framework for designing more trustworthy reinforcement learning systems and generating more robust system requirements.
Research Intern
Microsoft Research
· Redmond, WA
May 2025–Aug. 2025
  • Worked with the Software Analysis and Intelligence in Engineering Systems (SAINTES) group in MSR to automate differential testing for a product group.
  • Developed an AI agent for differential testing with traceability to key project artifacts, enabling engineers to validate outputs and ensure trustworthy results.
  • Ongoing research focuses on extending these methods beyond differential testing to broader software engineering tasks and product groups.
Co-founder | Lead Researcher
SAFA.ai
· Notre Dame, IN
Apr. 2022–Sep. 2024
  • Co-founded SAFA.ai, a platform leveraging AI to improve traceability and requirements management in safety-critical systems.
  • Secured an STTR Phase I grant with NASA and completed nine industry pilot projects.
  • Accepted into the 2023 Techstars Los Angeles Accelerator.
Software & Systems Engineer
Lockheed Martin
· Mount Laurel, NJ
Apr. 2019–Mar. 2021
  • Worked on the Aegis System at the Rotary and Mission Systems Branch.
  • Contributed to the development of a Java-based backend controller that reported real-time status of ship elements to user-facing interfaces.
  • Automated system-level tests with a Python framework and led refactoring to improve modularity and maintainability.

Publications & Writing

Blog Posts

1
Cross-Layer Transcoders are Incentivized to Learn Unfaithful Circuits
Georg Lange, RGRGRG, Kat Dearstyne, Kamal Maher
LessWrong
https://www.lesswrong.com/posts/6CS2NDmoLCFcEJMor/cross-layer-transcoders-are-incentivized-to-learn-unfaithful

Publications

1
Revealing the Dark Matter: Connecting Tacit and System Knowledge in Human-AI Collaborations
Katherine R. Dearstyne, Carmen Badea, Christian Bird, Robert DeLine
2026 ACM/IEEE International Conference on Software Engineering (ICSE), NIER Track, Rio de Janeiro, Brazil
https://cabird.com/pdfs/dearstyne2026revealing.pdf
2
Real-World Traceability Patterns for Generative AI Systems: With Insights from the Safa Dataset
Katherine R. Dearstyne, Alberto D. Rodriguez, Jane Cleland-Huang
2026 RAISE Workshop on Requirements Engineering for AI-powered SoftwarE at ICSE, Rio de Janeiro, Brazil
3
QuestRL: A Q&A Framework for Designing Trustworthy Reinforcement Learning Systems
Katherine R. Dearstyne, Pedro T. Alarcon Granadeno, Tyler Chambers, Jane Cleland-Huang
2025 IEEE 33rd International Requirements Engineering Conference, Valencia, Spain
https://ieeexplore.ieee.org/abstract/document/11190358
4
Assessing Compliance of Software System Designs to Laws, Regulations, and their Underlying Values
Aleksandra Marczak-Czajka, Katherine R. Dearstyne, Jane Cleland-Huang
2025 IEEE/ACM International Workshop on Designing Software (Designing) at ICSE, Ottawa, ON, Canada
https://ieeexplore.ieee.org/document/11029552
5
Supporting Software Maintenance with Dynamically Generated Document Hierarchies
Katherine R. Dearstyne, Alberto D. Rodriguez, Jane Cleland-Huang
2024 IEEE International Conference on Software Maintenance and Evolution, Flagstaff, AZ
https://ieeexplore.ieee.org/abstract/document/10795041
6
ROOT: Requirements Organization and Optimization Tool
Katherine R. Dearstyne, Alberto D. Rodriguez, Jane Cleland-Huang
2024 IEEE International Conference on Software Maintenance and Evolution, Flagstaff, AZ
https://ieeexplore.ieee.org/abstract/document/10795032
7
Prompts Matter: Insights and Strategies for Prompt Engineering in Automated Software Traceability
Alberto D. Rodriguez, Katherine R. Dearstyne, Jane Cleland-Huang
2023 IEEE International Requirements Engineering Conference Workshops, Hanover, Germany
https://ieeexplore.ieee.org/abstract/document/10260721
8
SAFA: A Tool for Supporting Safety Analysis in Evolving Software Systems
Alberto D. Rodriguez, Timothy Newman, Katherine R. Dearstyne, Jane Cleland-Huang
2022 IEEE/ACM International Conference on Automated Software Engineering, Rochester, MI
https://dl.acm.org/doi/10.1145/3551349.3559535
9
MEWS: Real-time Social Media Manipulation Detection and Analysis
Trenton W. Ford, Michael Yankoski, William Theisen, Tom Henry, Farah Khashman, Katherine R. Dearstyne, Tim Weninger, Pamela Bilo Thomas
2021 Conference on Neural Information Processing Systems (NeurIPS), Competition and Demonstration Track, Virtual
https://proceedings.mlr.press/v176/ford22a/ford22a.pdf