Fangyuan Xu
My name in Chinese: 许方园
Pronouns: she/her
📧: carrie[.]xfy[at]gmail.com
GithubSemantic Scholar
Google Scholar
CV(Feb 2026)
👩💻
I am a final-year Ph.D. student in Computer Science at New York University (Courant Institute), advised by Eunsol Choi.
I spent the first two years of my Ph.D. at the University of Texas at Austin before transferring to NYU in 2024 with my advisor. Previously, I graduated from Cornell University with a M.Eng in Computer Science and the University of Hong Kong with a B.Eng in Computer Science. I previously interned at Google (Summer 2025), Microsoft (Spring 2025) and Allen Institute of AI (Summer 2023). I worked at Twitter as a Machine Learning Engineer before my Ph.D.
🌟 I am on the industry job market - please email me if you think I could be a fit for your organization!
Research Interest
My interests lie at the intersection of natural language processing and machine learning. Generally, I am interested in:- Efficient inference methods: context compression (RECOMP) and selective attention (RefreshKV).
- Synthetic data generation: through self-evolving agentic pipeline (SAGE) and privacy-preserving RL (DP-RFT).
- Evaluation and analysis for realistic tasks involving: long-form outputs, human-model interaction and under-represented user groups.
Selected Publications
Please see my Google Scholar for the full list.
DP-RFT: Learning to Generate Synthetic Text via Differentially Private Reinforcement Fine-tuning, Arxiv 2026
Fangyuan Xu, Sihao Chen, Zinan Lin, Taiwei Shi, Sydney Graham, Pei Zhou, Mengting Wan, Alexander Stein, Virginia Estellers, Charles Chen, Tadas Baltrusaitis, Richard Speyer, Jennifer Neville, Eunsol Choi, Longqi Yang
SAGE: Steerable Agentic Data Generation for Deep Search with Execution Feedback, EACL 2026 Findings
Fangyuan Xu, Rujun Han, Yanfei Chen, Zifeng Wang, I-Hung Hsu, Jun Yan, Vishy Tirumalashetty, Eunsol Choi, Tomas Pfister, Chen-Yu Lee
[code]
RefreshKV: Updating Small KV Cache During Long-form Generation, ACL 2025
Fangyuan Xu, Tanya Goyal*, Eunsol Choi*
RECOMP: Improving Retrieval-Augmented LMs with Compression and Selective Augmentation, ICLR 2024
Fangyuan Xu, Weijia Shi, Eunsol Choi
[code]
KIWI: A Dataset of Knowledge-Intensive Writing Instructions for Answering Research Questions, ACL 2024 Findings
Fangyuan Xu, Kyle Lo, Luca Soldaini, Bailey Kuehl, Eunsol Choi, David Wadden
[Data]
Understanding Retrieval Augmentation for Long-Form Question Answering, COLM 2024
Hung-Ting Chen, Fangyuan Xu*, Shane Arora*, Eunsol Choi
[code]
Long-form Answers to Visual Question from Blind and Low Vision People, COLM 2024 (Spotlight)
Mina Huh, Fangyuan Xu, Yi-Hao Peng, Chongyan Chen, Hansika Murugu, Danna Gurari, Eunsol Choi, and Amy Pavel
[code]
A Critical Evaluation of Evaluations for Long-form Question Answering, ACL 2023
Fangyuan Xu*, Yixiao Song*, Mohit Iyyer, Eunsol Choi
[code]
How Do We Answer Complex Questions: Discourse Structure of Long-form Answers, ACL 2022
Fangyuan Xu, Jessy Junyi Li, Eunsol Choi
[code][website]
*=Equal contribution
Research Experience
Google Cloud AI Research, Summer 2025Mentor: Rujun Han
Microsoft Office of Applied Research, Spring 2025Mentor: Sihao Chen
Allen Institute of AI, Summer 2023Mentor: David Wadden
Last updated: February 2026
