About Me

I am Yingcheng Liu (刘应程), a first-year Ph.D. student at the School of Computer Science and Technology, Tongji University, Shanghai, China.

My research interests primarily include deep learning, reinforcement learning, multimodal large language models (MLLMs), and computer-using agents.

Outside of research, I enjoy playing basketball and swimming.

🛠 Technical Skills

• I am proficient in C++, Python, and Java, with hands-on development experience and the ability to confidently read, write, and maintain code and scripts.
• I am comfortable working in Linux-based environments for software development and research.
• I have a strong mathematical background and an in-depth understanding of the principles underlying deep learning and reinforcement learning.
• I have a solid understanding of mainstream LLM and VLM architectures and training paradigms, with practical experience using open-source model weights for inference, fine-tuning, and training.
• I am familiar with the fundamental architectures, components, and workflows of agentic systems.
• I scored 517 on CET-4 and 464 on CET-6 and can read academic literature and communicate effectively in English.

🔬 Research Experience

MoLE: Mixture of Latent Experts for Complementary Visual Reasoning

First Author · Under Review at ICLR 2027 · arXiv

I developed MoLE, a Mixture of Latent Experts framework that improves latent visual reasoning in vision-language models by encouraging different latent experts to extract complementary visual evidence. MoLE combines dynamic visual-token routing, expert-specific visual transformations, and latent summary experts within a two-stage training pipeline. Across five visual reasoning benchmarks, it achieves an average score of 78.6, outperforming data-matched supervised fine-tuning by 4.9 points and the strongest evaluated latent visual reasoning baseline under the same latent budget by 3.6 points.

🎖 Honors and Awards

• 2025–2026: Third Prize, Outstanding Undergraduate Student Scholarship, Tongji University.
• 2025: Second Prize, 2nd “SSE Cup · Gathering Talents in Shanghai” Innovation and Entrepreneurship Competition for University Students in Shanghai.
• 2025: Third Prize, 17th Shanghai College Students Computer Application Competition.
• 2022–2023: Second Prize, Outstanding Undergraduate Student Scholarship, Tongji University.

📖 Education

• September 2026 – Present: Ph.D. Student, School of Computer Science and Technology, Tongji University, Shanghai, China.
• September 2022 – June 2026: Undergraduate Student, School of Computer Science and Technology, Tongji University, Shanghai, China.
• September 2019 – June 2022: High School Student, Changsha No. 1 High School, Changsha, Hunan, China.