Zachary Yahn

CS PhD Student, Georgia Institute of Technology
AI Safety and Security · Computer Vision · Deep Learning

About

Hello! I'm Zach, a computer science PhD student at Georgia Tech. I'm advised by Prof. Ling Liu in the Distributed Data-Intensive Systems Lab (DiSL). My research spans deep learning for computer vision, trustworthy and secure ML, and AI safety, privacy, and data protection. I'm broadly interested in designing and evaluating safe and secure AI systems for public good.

My work has been recognized at NVIDIA's GTC keynote and by NASA's chief scientist, and I am the recipient of multiple nationally-competitive scholarships. I have also published several peer-reviewed articles in prestigious venues. I love meeting new people, and I'm always excited to work on new problems. Feel free to reach out if you'd like to connect!

Education

  • Ph.D. Computer Science – Georgia Institute of Technology (2024–Present)
  • M.S. Computer Science – University College Dublin (2023–2024), First-Class Honours
  • B.S. Computer Science – University of Virginia (2019-2023), Highest Honors

Awards & Honors

  • Presidential Fellow - Georgia Tech
  • George J. Mitchell Scholar
  • Outstanding Engineering Student - UVA
  • National Winner - iDISPLA Adversarial AI/ML Challenge
  • Rodman Scholar
  • Jefferson Scholar

Selected Publications

First Author

Adversarial attention attacks on large object detection transformers

Z. Yahn, S. Tekin, F. Ilhan, S. Hu, T. Huang, M. Loper, L. Liu
International Conference on Computer Vision (ICCV), 2025

Rapid automated detection of Titan clouds with instance segmentation

Z. Yahn, E. Duncan, D.M. Trent, J. Santerre, B. Seignovert, C.A. Nixon
Journal of Geophysical Research Machine Learning Edition, 2025

TinyNina: A lightweight super-resolution framework for satellite-based NO₂ prediction

Z. Yahn, P. Dey, B.S. Phelan, S. Dev
IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing (under review)

Titan cloud identification with deep transfer learning

Z. Yahn, D.M. Trent, E. Duncan, J. Santerre, C.A. Nixon
International Conference on Learning Representations (ICLR), 2023

Co-Author

A survey on large language model-based game agents

S. Hu, T. Huang, G. Liu, R.R. Kompella, F. Ilhan, S. Tekin, Y. Xu, Z. Yahn, L. Liu
arXiv 2025

Language-vision planner and executor for text-to-visual reasoning

Y. Xu, G. Liu, R. R. Kompella, S. Hu, T. Huang, F. Ilhan, Z. Yahn, L. Liu
Neural Information Processing Systems (NeurIPS), 2025 (under review)

Safety tax: Safety alignment makes your large reasoning models less reasonable

T. Huang, S. Hu, S. Tekin, Z. Yahn, Y. Xu, L. Liu
arXiv 2025

Multi-agent reinforcement learning with focal diversity optimization

S. Tekin, F. Ilhan, T. Huang, S. Hu, Z. Yahn, L. Liu
arXiv 2025

H^3 fusion: Helpful, harmless, honest fusion of aligned LLMs

S. Tekin, F. Ilhan, T. Huang, S. Hu, Z. Yahn, L. Liu
ACL ARR 2025 (under review)

Data poisoning and leakage analysis in federated learning

W. Wei, T. Huang, Z. Yahn, A. Singhal, M. Loper, L. Liu
Handbook of Trustworthy Federated Learning

Experience

Machine Learning Fellow

10a Labs
Apr. 2025 – Present
Contribute to client-facing projects in ML security and trustworthy AI for frontier systems. Responsible for production code deployed to cloud environments.

Graduate Research Assistant

Georgia Institute of Technology
Aug. 2024 – Present
Research adversarial face encryption for privacy protection in facial recognition and adversarial attacks on deep vision models.

Graduate Research Assistant

University College Dublin
Apr. 2024 – Aug. 2024
Developed TinyNina, an ultra-lightweight super-resolution model for satellite imagery with 1000× fewer parameters than SOTA models. Achieved top NO₂ prediction performance.

Machine Learning Research Intern

NASA Goddard Space Flight Center
Aug. 2021 – Aug. 2023
Created an instance segmentation approach for detecting methane clouds in Titan's atmosphere. Recognized by NVIDIA and NASA for outstanding AI applications.

NSF REU Research Fellow

Duke University
May 2021 – Aug. 2021
Constructed a classification and regression tree model to classify injuries on U.S. Navy aircraft carriers, including extensive data cleaning and feature selection.

Graduate Teaching Assistant

Georgia Institute of Technology
May 2025 – Aug. 2025
Coordinated teaching staff, held office hours, and graded assignments as head TA for CS6675: Internet Systems and Services

Let's Connect

Interested in collaboration or have questions about my research? Please reach out!