Tiansheng Hu

prof_pic.jpg

I’m a senior undergraduate at NYU Shanghai majoring in Computer Science. My research interests include AI for Scientific Research, Deep Research Agents, and Information Retrieval.

My research focuses on search agents that retrieve and reason over scientific literature: evaluating existing systems (SciArena), improving retrieval for deep research agents (SAGE), and training page-native agents that read documents directly as page images instead of OCR text (DR Pixel). I am also interested in making agentic systems reliable in high-stakes domains such as finance (FinTrust, FinLFQA).

I am fortunate to work with Prof. Chen Zhao (NYU Shanghai), Prof. Arman Cohan (Yale University), Prof. Sewon Min (UC Berkeley), and Yilun Zhao (Yale University). In Summer 2026, I was a visiting student researcher at Berkeley NLP, advised by Prof. Sewon Min.

news

Nov 02, 2026 Released DR Pixel, an open-source page-native deep research agent that reads scientific papers directly as page images, with no OCR. This work was done during my summer at Berkeley NLP with Prof. Sewon Min. [Code]
Jun 01, 2026 Started my summer visit at Berkeley NLP as a visiting student researcher, advised by Prof. Sewon Min!
Aug 20, 2025 Two papers accepted at EMNLP 2025! FinTrust at Main and FinLFQA at Findings.
Aug 15, 2025 SciArena accepted at NeurIPS 2025 Datasets and Benchmarks Track as a Spotlight!

selected publications

  1. Preprint
    drpixel.jpg
    DR Pixel: Page-Native Deep Research with Vision-Language Models
    Tiansheng Hu, Yichuan Wang, James V. Talwar, and 3 more authors
    2026
  2. Preprint
    sage.jpg
    SAGE: Benchmarking and Improving Retrieval for Deep Research Agents
    Tiansheng Hu, Yilun Zhao, Canyu Zhang, and 2 more authors
    2026
  3. NeurIPS
    sciarena.jpg
    SciArena: An Open Evaluation Platform for Non-Verifiable Scientific Literature-Grounded Tasks
    Yilun Zhao*, Kaiyan Zhang*, Tiansheng Hu*, and 15 more authors
    In NeurIPS 2025 Benchmark and Dataset Track (Spotlight), 2025
  4. EMNLP
    fintrust.png
    FinTrust: A Comprehensive Benchmark of Trustworthiness Evaluation in Finance Domain
    Tiansheng Hu, Tongyan Hu, Liuyang Bai, and 3 more authors
    In EMNLP 2025 Main, 2025
  5. EMNLP
    finlfqa.jpg
    FinLFQA: Evaluating Attributed Text Generation of LLMs in Financial Long-Form Question Answering
    Yitao Long*, Tiansheng Hu*, Yilun Zhao, and 2 more authors
    In Findings of EMNLP 2025, 2025