Hello, I’m

Penghui Yang 杨鹏晖

Ph.D. Student · Multimodal AI Researcher

I study how multimodal foundation models learn, reason, and align. I am a Ph.D. student at Shanghai AI Lab, jointly trained with the School of Artificial Intelligence, Shanghai Jiao Tong University, under the supervision of Yuhang Zang.

Multimodal Learning Vision–Language Models Post-training Alignment
View my research CV
Penghui Yang
Based in Shanghai, China

Research focus

Building capable, efficient, and trustworthy multimodal systems.

My work sits at the intersection of representation learning, model efficiency, and alignment.

01

Multimodal Learning

Learning unified representations across vision, language, and diverse real-world signals.

02

Vision–Language Models

Improving visual understanding through scalable pre-training and effective data design.

03

LLM Post-training/Alignment

Aligning foundation models with reliable rewards, preferences, and human intent.

CapRL++ method comparison diagram

Featured work · 2026

A unified framework for dense visual captioning and multimodal post-training.

Our latest work explores a scalable approach to stronger perception and alignment in multimodal large language models.

TPAMI Submission Multimodal LLMs
Read the paper

Updates

Latest news

  1. Received the Beijing Outstanding Graduate and Tsinghua Excellent Graduate awards.

  2. Our work CapRL++ was submitted to TPAMI.

  3. Started my internship at Shanghai AI Lab

  4. Received the National Scholarship.

  5. Won Second Prize in Tsinghua’s AGI Competition for a RAG project.

Background

Education & experience

  1. Ph.D. Student

    Shanghai AI Lab × Shanghai Jiao Tong University

  2. Research Intern

    Shanghai Artificial Intelligence Laboratory

  3. B.Eng. · Rank 1/16

    Building Environment & Energy Engineering, Tsinghua University

Recognition

Honors & awards

2026Beijing Outstanding GraduateBeijing Municipal Education Commission
2026Tsinghua Excellent GraduateTsinghua University
2025National ScholarshipMinistry of Education of China
2025Second Prize · AGI CompetitionTsinghua University
2023–24Comprehensive Excellence ScholarshipTop 10% · Tsinghua University

Beyond the lab

Good ideas often come after a workout—or with good coffee.

Outside research, you’ll find me playing basketball or billiards, training at the gym, and exploring cafés around the city.