Andong Hua

PhD student @ UC Santa Barbara

I am a fourth-year PhD student in the Electrical and Computer Engineering (ECE) Department at UCSB, advised by Prof. Yao Qin. I am also a member of the REAL AI Lab. I have worked as an Applied Scientist Intern at Amazon during the summers of 2025 and 2026, on Agentic RL credit assignment and prompt optimization. Previously, I worked as a Research Engineer at TuSimple, focusing on developing perception systems for autonomous driving trucks. Prior to that, I obtained my Master's degree in Electrical and Computer Engineering from UCLA.

My research interests broadly lie in the areas of machine learning and artificial intelligence, with a focus on:

  • Robustness and safety in large language models, vision models, and multimodal models.
  • AI for healthcare, such as nutrition estimation.

NEWS!

  • Sep. 2026: TokenSwap, my first author paper on the image-text modality gap in multimodal LLMs done in collaboration with Google DeepMind, has been accepted to NeurIPS 2026!
  • June. 2026: Excited to join the AWS Fundamental Research Team as an Applied Scientist Intern, focusing on Agentic RL credit assignment research!
  • Aug. 2025: My co-first author paper on prompt sensitivity in LLM evaluation has been accepted to EMNLP 2025 (Main)
  • June. 2025: Excited to join the Amazon Smart Vehicle Team as an Applied Scientist Intern, focusing on prompt optimization for function calling!
  • Jan. 2025: My co-first author paper on benchmarking LLMs for nutrition estimation from meal descriptions has been accepted to ICLR 2025!
  • Feb. 2024: First author paper on adversarial transfer learning is accepted to CVPR 2024.

Selected Publications

TokenSwap: Benchmarking and Reducing the Modality Gap in Multimodal LLMs.
Andong Hua, Colton Bishop, Igor Mordatch, Arian Hosseini, Jindong Gu, Aleksandra Faust, Rebecca Roelofs, Yao Qin
Conference on Neural Information Processing Systems (NeurIPS), 2026.
[Paper]
Flaw or Artifact? Rethinking Prompt Sensitivity in Evaluating LLMs.
Andong Hua*, Kenan Tang*, Chenhe Gu, Jindong Gu, Eric Wong, Yao Qin
Empirical Methods in Natural Language Processing (EMNLP, Main Conference), 2025.
[Paper]
NutriBench: A Dataset for Evaluating Large Language Models in Nutrition Estimation from Meal Descriptions.
Andong Hua*, Mehak Preet Dhaliwal*, Laya Pullela, Ryan Burke, Yao Qin
International Conference on Learning Representations (ICLR), 2025.
[Paper] [Project Page] [Data]
Improving Adversarial Transferability in MLLMs via Dynamic Vision-Language Alignment Attack.
Chenhe Gu, Jindong Gu, Andong Hua, Yao Qin
Preprint, 2024.
[Paper]
Initialization Matters for Adversarial Transfer Learning.
Andong Hua, Jindong Gu, Zhiyu Xue, Nicholas Carlini, Eric Wong, Yao Qin
Computer Vision and Pattern Recognition (CVPR), 2024.
[Paper] [Code]

Experience

University of California, Santa Barbara

PhD Student, Department of Electrical and Computer Engineering
Jan 2024 – Present

Amazon

Applied Scientist Intern, AWS Fundamental Research Team
June 2026 – Sep 2026

Amazon

Applied Scientist Intern, Amazon Smart Vehicle Team
June 2025 – Sep 2025

University of California, Santa Barbara

Research Assistant (RA)
Feb 2023 – Jan 2024

TuSimple

Research Engineer, Perception Team
Jun 2022 – Feb 2023

University of California, Los Angeles

Master Student, Department of Electrical and Computer Engineering
Sep 2020 – Jun 2022

University of Nottingham

Bachelor Student, Department of Electrical and Electronic Engineering
Sep 2016 – Jun 2020