Jingwen Yang

Jingwen Yang

The Hong Kong Polytechnic University

✨ Welcome to My Website!

Hi there! 😊 I’m Jingwen Yang (Anita), a third-year undergraduate at The Hong Kong Polytechnic University, majoring in Data Science & Analytics in the Department of Applied Mathematics and minoring in Computer Science in the Department of Computing. I’m deeply grateful to be advised by Prof. Wanyu Lin, Prof. Hongxia Yang and Prof. Manling Li in my research journey.

🔬 Research Interests

I’m currently working on projects related to spatial intelligence of VLMs. Broadly, my research interests lie in multimodal reasoning, embodied agents, and human-centered NLP. I’m especially excited about building AI systems that can reason, generalize, and interact robustly in real-world settings, with a focus on explainability, value alignment, and cognitive grounding.

I’m actively seeking Summer 2026 research internship opportunities related to NLP, multimodal learning, and trustworthy AI. I’d truly love the opportunity to connect, contribute, and grow! 🌷

📩 Get in Touch

I’m genuinely thankful for the mentorship and support I’ve received so far, and always happy to connect with others interested in the NLP field 🤍
If you’re open to chatting or collaboration, feel free to contact me via email 📧 jingwen.yang@connect.polyu.hk or add me on WeChat 💬 jwabo001 — I’d truly appreciate the opportunity！

📝 Publications & Preprints

† indicates corresponding author, * indicates equal contribution

arXiv preprint
Infi-Med teaser

🌟 Infi-Med: Low-Resource Medical MLLMs with Robust Reasoning Evaluation

Zeyu Liu*, Zhitian Hou*, Yining Di, Kejing Yang, Zhijie Sang, Congkai Xie, Jingwen Yang, Siyuan Liu, Jialu Wang, Chunming Li, Ming Li, Hongxia Yang †

• arXiv preprint, 2025    📄 arXiv

A robust and resource-efficient medical MLLM, systematically evaluated across reasoning, grounding, and hallucination dimensions on real-world clinical VQA datasets.

arXiv preprint
Crystal Canonicalization teaser

🔷 Revisiting the Canonicalization for Fast and Accurate Crystal Tensor Property Prediction

Haowei Hua, Jingwen Yang, Wanyu Lin†, Pan Zhou

• AAAI-2026 (Oral)    📄 arXiv

A polar decomposition-based canonicalization framework (GoeCTP) for O(3)-equivariant crystal tensor property prediction, achieving up to 13× speedup and SOTA accuracy.

💻 Internships

  • Jun 2025 – Aug 2025 · R&D Intern, Nokia
    Shanghai, China

📖 Education

  • Jan 2026 - May 2026 · Semester Visiting
    University of California, Berkeley

  • Jun 2024 – Aug 2024 · Summer Session
    Stanford University

  • Sep 2023 – Jun 2027 (expected) · BSc (Hons) in Data Science & Analytics, Minor in Computer Science
    The Hong Kong Polytechnic University