Yanqi Dai
代彦琪

Expected to graduate in June 2027, I am seeking industry positions focused on foundation models, including LLMs, VLMs, and UMMs. Feel free to reach out if you have relevant opportunities or insights to share.

About me

Capable and versatile multimodal intelligence.

I am a fourth-year Ph.D. candidate in the Gaoling School of Artificial Intelligence at Renmin University of China, advised by Prof. Zhiwu Lu. I was a visiting student in the College of Computing and Data Science at Nanyang Technological University, advised by Prof. Hanwang Zhang. Before my doctoral studies, I received my B.E. from the School of Software at Dalian University of Technology.

My research and industry experience spans large multimodal models, reinforcement learning for reasoning, and multi-task optimization, covering the full R&D pipeline from data construction and filtering to algorithm design, model training, and evaluation. I have published multiple first-author papers at ICLR, WWW, and UAI, with the broader goal of building more capable and versatile multimodal models. In industry, I work extensively on data construction and filtering for vision-language pretraining of Zhipu AI's GLM foundation models. Previously, as a core technical contributor on the startup team at Metabrain AGI, I helped develop ChatImg and the Awaker series of Chinese large multimodal models.

Large
Multimodal
Models
Data & Task Balancing
RL for Reasoning
Multimodal Role-Playing Agents
Visual Instruction Tuning
Selected papers

Research highlights.

ICLR 2026MathForge

Harder Is Better: Boosting Mathematical Reasoning via Difficulty-Aware GRPO and Multi-Aspect Question Reformulation

Yanqi Dai, Yuxiang Ji, Xiao Zhang, Yong Wang*, Xiangxiang Chu, Zhiwu Lu*

Hugging Face Daily Papers #2 of the day.

WWW 2026 · OralVisATB

Adaptive Task Balancing for Visual Instruction Tuning via Inter-Task Contribution and Intra-Task Difficulty

Yanqi Dai, Yong Wang, Zebin You, Dong Jing, Xiangxiang Chu, Zhiwu Lu*

ICLR 2025MMRole

MMRole: A Comprehensive Framework for Developing and Evaluating Multimodal Role-Playing Agents

Yanqi Dai, Huanran Hu, Lei Wang, Shengjie Jin, Xu Chen*, Zhiwu Lu*

Technical ReportAwaker2.5-VL

Awaker2.5-VL: Stably Scaling MLLMs with Parameter-Efficient Mixture of Experts

Jinqiang Long†, Yanqi Dai†, Guoxing Yang, Hongpeng Lin, Nanyi Fei, Yizhao Gao, Zhiwu Lu

UAI 2023IGB

Improvable Gap Balancing for Multi-Task Learning

Yanqi Dai, Nanyi Fei, Zhiwu Lu*

† Equal contribution. * Corresponding author.

Experience

From research labs to foundation-model teams.

May 2026 — Present · Beijing

Algorithm Intern, VLM Pretraining Team in Base Group · Zhipu AI

Mar 2025 — Sep 2025 · Beijing

Research Intern, Machine Learning Team · Amap-Alibaba Group

Jun 2023 — Feb 2025 · Beijing

Algorithm Intern, Model Team · Metabrain AGI

Selected honors & awards

Recognition.

Excellent Scholarship for Postgraduate StudyRenmin University of China · 2025
Excellent Scholarship for Postgraduate StudyRenmin University of China · 2023
Scientific Research FundRenmin University of China · 2023
First-Class Academic Scholarship for Postgraduate StudyRenmin University of China · 2023
Outstanding GraduateLiaoning Province · 2022
Toly Bread Alumni ScholarshipDalian University of Technology · 2020 / 2019
First Prize National Mathematics Competition for College Students2019
Blogs & notes

Learning, distilled.