A path shaped
by curiosity.

Researcher

Shanghai AI Laboratory, Shanghai, China

• Conduct research in AI for Science, focusing on multimodal models and scientific reasoning.
• Investigate equation discovery, visual-to-symbolic reasoning, and structural inference of dynamical systems.
• Contribute to scientific foundation models, native structural reasoning, and physics reasoning benchmarks.

Graduate Assistant for Autonomous Driving

Karlsruhe Institute of Technology, Germany

• Engineered an innovative visual localization technique using Graph Neural Networks and OpenCV to enhance the precision of monocular camera-based navigation in autonomous vehicles.
• Successfully presented and published the outcomes of the visual localization research in a peer-reviewed conference.
• Investigated combinatorial approaches to hierarchical visual localization, broadening the scope of research and uncovering potential advancements in the domain.

Research Intern

Robert Bosch GmbH, Renningen, Germany

• Organized and delivered a tutorial on AUTOSAR for the research team, initiated communication with the director.
• Pioneered the development of a diagnosis system using an extended Kalman filter for an autonomous electric vehicle prototype’s electric propulsion system.
• Explored the integration of modern AI methods into self-driving cars for diagnostic purposes.

PhD in Computer Science

Faculty of Science, Technology and Medicine, University of Luxembourg, Luxembourg

• Structural Inference for Dynamical Systems
• Network Science
• Graph Theory
• Supervised by Prof. Jun Pang

Master of Science (MSc)

Karlsruhe Institute of Technology, Germany

• Cognitive Systems
• Autonomous Driving

Bachelor of Engineering (BEng)

Tongji University, China

• Mechanics and Dynamics
• Vehicle Engineering
• Graduated with Honors

Multimodal Models and Scientific Reasoning

• Research on scientific multimodal foundation models and native structural reasoning across scientific domains.
• Visual-to-symbolic discovery of physical laws with VIPER-R1.
• Physics reasoning evaluation with PhysUniBench and scientific instruction following with SciIF.

Structural Inference (AI4Science)

• Structural inference of dynamical systems from observational trajectories.
• Methods that handle irregular sampling, incomplete observations, and partial prior knowledge.
• Developed StructInfer, an open benchmark for evaluating structural inference methods across disciplines.
• Research includes SICSM, SGSI, and generative flow network approaches to structural inference.

AI4EDU

• Developed models for recommendation systems using spatio-temporal graphs, enhancing them with integrated database solutions.
• Innovated techniques for addressing missing data issues by leveraging spatio-temporal similarities.
• Launched the LLM4EDUKG benchmark, assessing the reasoning capabilities of pre-trained large language models on domain-specific knowledge graphs.

• ICML2023 Student Travel Grant

• NeurIPS2022 Student Travel Grant

• Bachelor Graduation with Honors