Last update: 2026-06-22

Hi there! I’m Shiyan Liu (刘师言), a recent BEng graduate in Data Science & Big Data Technology from Huazhong University of Science and Technology (HUST). I spent a wonderful Spring 2026 semester at UC Berkeley as a visiting student, and I’m now heading to Imperial College London for an MSc in Computing (AI&ML) in Fall 2026. I also had a fulfilling internship at JD.com in 2025, and am glad to be back there this summer as a Machine Learning Engineer.

My research interests include machine learning, deep learning, reinforcement learning, and data science. My prior reseach experiences focused on deep reinforcement learning and agentic AI systems. I look forward to collaborating with fellow researchers and contributing to the frontiers of AI. None of this would have been possible without the people who believed in me along the way. → Acknowledgements

Outside of research, I love playing football and have proudly been a devoted Tottenham Hotspur supporter since 2018. Come on you Spurs! 🤍⚽️

🔬 Research Interests Finding the right answer — and knowing it's right — across graphs, documents, and language.

Neural Combinatorial Optimization	Geometric equivariance: MViewRouter · Multimodal fusion: VAGPO
Dense Retrieval	Positive-unlabeled: PURE · Test-time adaptation: DART
Trustworthy LLM Systems	Probabilistic evaluation: DICE · Explainable optimization: VISTA

I am currently and actively looking for (Volunteer) Research Assistant opportunities, and would greatly appreciate any leads.
Please feel free to reach me at shyl@hust.edu.cn (institutional) or shyliu.china@gmail.com (personal).

📖 Educations

Imperial College London Prospective
MSc Computing (AI&ML) · 2026.09 - 2027.12 · London, UK

UC Berkeley
Visiting Student · 2026.01 - 2026.05 · Berkeley, CA, USA

Huazhong University of Science and Technology (HUST)
BEng Data Science & Big Data Technology · 2022.09 - 2026.06 · Wuhan, China

💻 Internships

JD.com Ongoing
Machine Learning Engineer · 2026.06 - now · Beijing, China

Reinforcement Learning & Intelligent Computing Group, HUST
Research Assistant · Fortunately advised by Prof. Yan Jin · 2024.06 - now · Wuhan, China

JD.com
Software Development Engineer · 2025.03 - 2025.08 · Beijing, China

📝 Publications & Preprints

[2026.04] Liu, S., Li, Y. Test-Time Training for Zero-Resource Dense Retrieval Reranking. Accepted @ KnowFM · ACL 2026 [paper]
[2026.04] Liu, S., Xia, Q., Xia, Q., Liu, Y., Yu, X. & Qu, R. Reflection in the Dark: Exposing and Escaping the Black Box in Reflective Prompt Optimization. Accepted @ ACL 2026 SRW [paper]
[2026] Liu, S., Tan, B., Wu, Y., & Jin, Y. MViewRouter: Internalizing Geometric Equivariance via Multi-view Alternating Attention for Combinatorial Routing. Under Review [paper]
[2025.11] Liu, S., Ma, J., & Qu, R. DICE: Discrete Interpretable Comparative Evaluation with Probabilistic Scoring for Retrieval-Augmented Generation. Accepted @ ResponsibleFM · NeurIPS 2025 [paper] [code]
[2025] Liu, S., Tan, B., Cao, Z., & Jin, Y. VAGPO: Vision-augmented Asymmetric Group Preference Optimization for Graph Routing Problems. Under Review [paper]
[2025] Liu, S., Qu, R., & Jin, Y. FluentLip: A Phonemes-Based Two-stage Approach for Audio-Driven Lip Synthesis with Optical Flow Consistency. arXiv Preprint [paper]

🎖 Honors and Awards

2026.04 Outstanding Graduate (HUST).
2025.09 National Scholarship.
2025.09 Outstanding Student Scholarship (HUST).
2025.05 Tencent Scholarship.
2024.09 Academic Excellence Scholarship (HUST).