Daiqing Qi taking a mirror photo with a Leica camera

Daiqing Qi

Microsoft

I am building Copilot at Microsoft. I am broadly interested in multimodal foundation models (MLLMs, VLMs) and LLM agents.

I received my PhD from UVA, working with Dr. Sheng Li. I also spent time at Adobe Research and Amazon, working on multimodal learning and LLMs. I co-organized workshops on Generative AI at WACV’26 and NeurIPS’24.

Here are some photographs and interesting stuff.

Publications

  1. PixelDense: Dense Prediction as Representation Alignment for Pixel Diffusion

    L. Yang, Daiqing Qi, W. Zhang, A. Li, Y. Yang, Y. Li, Y. Kong, H. Zheng, Z. Zhang, Z. Lin, V. Jampani, S. Li

    NeurIPS 2026

    PaperProjectCode

  2. VITAL: Visual-Text Invariance Tests for Text-Prior Reliance in Multimodal Language Models

    P. Wang, Daiqing Qi, T. Wang, W. Zhang, L. Yang, S. Li

    Under Review

  3. Adaptive Content-Contrastive Decoding for Representation-Robust Tabular Reasoning

    H. Yuan, M. Hu, Daiqing Qi, V. Datla, A. Samuel, S. Li

    NeurIPS 2026 Workshop LP4FM(Oral)

  4. HFS: Holistic Query-Aware Frame Selection for Efficient Video Understanding

    Y. Yang, Y. Li, Daiqing Qi, L. Yang, T. Wang, W. Zhang, S. Li, K. Lam

    ACM MM 2026

    Paper

  5. Improve Temporal Reasoning in MLLMs via Video Contrastive Decoding

    Daiqing Qi, D. Guo, H. Yuan, H. Zhao, M. Hu, L. Yang, S. Li

    NeurIPS 2025

    Paper

  6. VRMDiff: Text-Guided Video Referring Matting Generation of Diffusion

    L. Yang, J. Song, T. Wang, Daiqing Qi, W. Shi, Y. Liu, S. Li

    arXiv 2025

  7. The Photographer’s Eye: Teach MLLMs to See and Critique like Photographers

    Daiqing Qi, H. Zhao, J. Shi, S. Jenni, Y. Fan, F. Dernoncourt, S. Cohen, S. Li

    CVPR 2025

    PaperCode

  8. No Free Lunch: Retrieval-Augmented Generation Undermines Fairness in LLMs, Even for Vigilant Users

    M. Hu, H. Wu, R. Zhu, Z. Guan, D. Guo, Daiqing Qi, S. Li

    EMNLP 2025 (Findings)

    Paper

  9. Video-LMM Post-Training: A Deep Dive into Video Reasoning with Large Multimodal Models

    Y. Tang, J. Bi, …, Daiqing Qi, …, C. Xu

    arXiv 2025

    PaperGitHub

  10. The Clever Hans Mirage: A Comprehensive Survey on Spurious Correlations in Machine Learning

    W. Ye, L. Jiang, …, Daiqing Qi, …, A. Zhang

    TMLR 2025

    Paper

  11. Easy Regional Contrastive Learning of Expressive Fashion Representations

    Daiqing Qi, H. Zhao, S. Li

    NeurIPS 2024

    PaperCode

  12. Generalizing (CLIP) to Unseen Domains via Text-guided Augmentation

    Daiqing Qi, H. Zhao, A. Zhang, S. Li

    ECCV 2024

    Paper

  13. Tag-grounded Visual Instruction Tuning with Retrieval Augmentation

    Daiqing Qi, H. Zhao, Z. Wei, S. Li

    EMNLP 2024 (Main Track)

    Paper

  14. Revealing an Overlooked Challenge in Class-Incremental Graph Learning

    Daiqing Qi, H. Zhao, X. Jia, S. Li

    TMLR 2024

    Paper

  15. A Survey of Trustworthy Representation Learning across Domains

    R. Zhu, D. Guo, Daiqing Qi, Z. Chu, X. Yu, S. Li

    TKDD 2024

    Paper

  16. Better Generative Replay for Continual Federated Learning

    Daiqing Qi, H. Zhao, S. Li

    ICLR 2023

    PaperCode

Awards

Workshops

Service

Organizer / Program Chair:

Reviewer / PC Member:

  • Neural Information Processing Systems (NeurIPS)
  • International Conference on Learning Representations (ICLR)
  • International Conference on Machine Learning (ICML)
  • IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)
  • European Conference on Computer Vision (ECCV)
  • Transactions on Machine Learning Research (TMLR)

Teaching:

  • DS 4002: Data Science Project Course (Spring 2023)
  • CS 5012: Foundations of Computer Science (Fall 2022)

Experience

Misc

I used to play the Auto Battler game and was ranked #52 among the top 100 best players nationwide.

Links:

A concrete studio overlooking a green, wooded valley
Forest Office