Tao Sun
Hi! I am an Applied Scientist at Amazon. My work focuses on AI agents, multi-modal models, in particular for shopping scenarios.
Previously, I did my PhD study in Computer Science at Stony Brook University, fortunately advised by Prof. Haibin Ling.
I am broadly interested in machine learning and computer vision, particularly AI agents, vision-language learning, domain adaptation, weakly-supervised learning. I also work on adversarial attack and defense of neural networks.
Contact: taosun.ai # gmail.com
News
| Apr 2025 | Our project Buy for Me shopping agent was launched! |
|---|---|
| Jan 2025 | Presented intern work at EvalMG25 @ COLING workshop. We worked on multi-modal In-Context-Learning for LLaVA model. |
| Nov 2024 | Started working as Applied Scientist at Amazon, Seattle! |
| Nov 2024 | Successfully defended my PhD thesis. |
| Oct 2024 | One paper on Semi-Supervised Learning was accepted by WACV 2025. |
Selected publications
-
LLaVA-RE: Binary image-text relevancy evaluation with multimodal large language model
Tao Sun, Oliver Liu, JinJin Li, Lan Ma
EvalMG 2025 | Paper • arXiv
-
Mask and Restore: Blind Backdoor Defense at Test Time with Masked Autoencoder
Tao Sun, Lu Pang, Chao Chen, Weimin Lyu, and Haibin Ling
arXiv 2023 | arXiv • Code
-
Local Context-Aware Active Domain Adaptation
Tao Sun, Cheng Lu, and Haibin Ling
ICCV 2023 | Paper • arXiv • Code
-
Backdoor Cleansing with Unlabeled Data
Lu Pang, Tao Sun, Haibin Ling, and Chao Chen
CVPR 2023 | Paper • arXiv • Code
-
Domain Adaptation with Adversarial Training on Penultimate Activations
Tao Sun, Cheng Lu, and Haibin Ling
AAAI 2023 | Oral Presentation • Paper • arXiv • Code
Industry experience
Applied Scientist, Amazon Shopping (Nov. 2024 - )
Applied Scientist Intern, Amazon Shopping (May 2024 - Aug. 2024)Multimodal LLM, Instruction tuning, Agents
Machine Learning Engineer Intern, Adobe DX (May 2023 - Aug. 2023)
Time series forecasting
Machine Learning Engineer, Ant Group (July 2018 - July 2019)
Recommender system
Software Engineer Intern, Baidu NLP (Sept. 2017 - Dec. 2017)
News matching