Portrait of Stuart Aldrich

Stuart Aldrich

PhD Student @ WashU • Studying LLM Behavior • Robotics & AI

About Me

I am a second year PhD student in Dr. Umar Iqbal's lab at Washington University in St. Louis. My current research is focused on studying LLM behavior. I am additionally broadly interested in robotics and artificial intelligence, and how we can make more intelligent machines to help us solve real-world problems.

Outside of research, I have strong personal interests in robotics and game development, and enjoy working on those topics in my spare time.

Email: aldrich.s@wustl.edu — aldrich.s@wustl.edu

Publications

Comprehensive Assessment of OCR Tools for Gene Name Recognition in Biological Pathway Figures
S. Aldrich, M. O. Arowolo, F. He, M. Popescu and D. Xu. 2022 IEEE International Conference on Bioinformatics and Biomedicine (BIBM), Las Vegas, NV, USA, 2022, pp. 3574–3579. DOI: 10.1109/BIBM55620.2022.9995448.

Keywords: Visualization; Image recognition; Sensitivity; Limiting; Text recognition; Biological system modeling; Optical character recognition; OCR; Gene; Biological Pathway; Biomedical Literature Mining; Name Recognition

View on IEEE Xplore

Blurb: Biological pathway figures hold valuable scientific insights, but their complexity makes it difficult for computers to extract the information they contain. We tested leading Optical Character Recognition (OCR) tools to determine which could most accurately identify gene names in these diagrams. Across multiple datasets, Google Cloud Vision and MMOCR emerged as the top performers.

Projects

roadtripevs.com

Your Practical Guide to Electric Road Trips

Visit site

RoadtripEVs is a personal, data-driven resource for finding and comparing EVs optimized for long-distance travel. It estimates charging time as a share of trip duration (10–80% charge) and factors in real-world electricity and gas costs for practical context. While built around my own EV search, it’s designed to help any driver make informed road trip choices.

commacars.com

Exploring Cars That Work with openpilot

Visit site

CommaCars is a reference site built around the openpilot comma 3x ecosystem. Instead of focusing on every vehicle on the market, it narrows in on those officially supported by Comma.ai’s driver-assistance platform, making it easier to see what’s realistically compatible today. The site also surfaces estimated resale pricing to give shoppers a ballpark sense of cost, with clear reminders to verify numbers through sources like Edmunds, KBB, NADA, or TrueCar. The goal isn’t to replace deep research, but to give drivers a quick starting point when considering whether their next car could work with openpilot.