Yuandong Pan
Postdoctoral Researcher, School of Engineering, Stanford University
I am a postdoctoral researcher developing AI for the digital transformation and lifecycle management of built environment. My work links large language models and vision-language models with geometry, asset relationships, and source evidence, so engineers can make verifiable decisions about buildings, roads, and other assets across their whole lifecycle.
About
I am a postdoctoral researcher in the School of Engineering at Stanford University, working with Prof. Michael Lepech. Before Stanford, I was a Marie Skłodowska-Curie Actions Future Road Research Fellow at the University of Cambridge with Prof. Ioannis Brilakis, and a Visiting Fellow at the University of Hong Kong. I completed my PhD at the Technical University of Munich, co-advised by Prof. André Borrmann and Prof. Ioannis Brilakis, on creating information-rich digital twins of indoor environments from image and point-cloud data.
My work takes a systems-and-deployment perspective: making learning methods reliable, efficient, and trustworthy in the real world, not only in principle. It has been recognised with prizes in the CVPR Scan-to-BIM competitions and a journal Best Paper Award, and supported by competitive fellowships and funding as principal investigator. I am also interested in translating research into practice: I completed the Stanford Ignite programme at the Graduate School of Business and co-founded a startup that uses LLMs to support logistics optimization.
Research interests
- Machine learning
- Large language models and multi-agent systems
- Multi-modal learning and perception
- Computer vision
- Small, specialised, and verifiable models
- Digital twins of buildings and infrastructure
- Automated inspection and condition assessment
- AI for engineering systems
News
- Sep 2026
Our paper on anchoring LLM-based chest X-ray report generation via diffusion language planning is accepted at NeurIPS 2026.
- Aug 2026
As first author: “Image-Guided Pavement Defect Recognition in GPR Data with Novel 3D Deep Learning Architecture” is accepted at the Journal of Computing in Civil Engineering.
- 2026
“EGT-KG: Evidence-Grounded Typed KG Retrieval for Practical Scientific QA with Small Language Models” is accepted at EMNLP 2026.
- 2026
“LLM-guided Semantic Bootstrapping for Interpretable Text Classification with Tsetlin Machines” appears in Findings of ACL 2026.
- 2026
As first author: “LLM-enabled Multi-agent Framework for Automated Scan-to-BIM and Scan-to-Graph Reconstruction” is published in Automation in Construction.
- 2026
As first author: “LLM-enabled Multi-agent Framework for Natural Language Interaction with Graph-Based Digital Twins” is published in Automation in Construction.
- 2026
As corresponding author: “Automating Graph-Based Geometric Digital Model Generation for Building Digital Twin Applications from Point Cloud and Image Data” is published in Building and Environment.
- 2026
As corresponding author: “Regional Graph-Based Pavement Condition Data Quality Assessment in Support of Trustworthy Highway Infrastructure Digital Twins” is published in Automation in Construction.
- 2026
“SemanticBridge: A Dataset for 3D Semantic Segmentation of Bridges and Domain Gap Analysis” is published in Developments in the Built Environment.
- 2026
“A Digital Twin-Based Approach for Dynamic Traffic-Aware Routing and Charging of Electric Vehicles” is published in Expert Systems with Applications.
- 2026
“Data Sharing Mechanism among Intelligent Transportation Infrastructure Stakeholders: Based on the Data Value Chain Perspective” is published in the Journal of Construction Engineering and Management.
- 2026
As first author: “Visual Language Models for Automated Road Defect Detection and Classification” appears at the FUTUREROADS Final Conference.
- 2026
“Perceiving Creativity in the Age of AI: How Labels, Beliefs, and Familiarity Shape Evaluations of AI-Generated and Human-Created Art” appears at the SocialLLM Workshop, ICWSM 2026.
- 2026
Completed Stanford Ignite, the Graduate School of Business certificate programme in innovation and entrepreneurship, and gave a guest lecture on digitization and AI in the built environment for CEE 121/221 at Stanford.
- 2025
Our paper on a trustworthy AI-supported digital twin framework for road operation and maintenance received the Smart Construction Best Paper Award for 2025. I also joined the journal's Youth Editorial Board.
- Sep 2025
Joined Stanford University as a postdoctoral researcher in the School of Engineering.
Publications
Experience and education
Appointments
- 2025–present
Postdoctoral ResearcherSchool of Engineering, Stanford University
- 2023–2025
Marie Skłodowska-Curie Actions Future Road Research FellowUniversity of Cambridge
- 2024
Visiting FellowUniversity of Hong Kong, Sino-British Fellowship Trust
- 2017–2018
Intern, Research and DevelopmentVolkswagen Group, Wolfsburg, Germany
Education
- 2026
Stanford Ignite, Certificate in Innovation and EntrepreneurshipStanford University Graduate School of Business
- 2019–2023
Ph.D.Technical University of Munich. Co-advised by André Borrmann and Ioannis Brilakis
- 2020–2021
Visiting Student, Department of EngineeringUniversity of Cambridge
- 2015–2019
Master's degree, Mechanical EngineeringKarlsruhe Institute of Technology
- 2011–2015
Bachelor's degree, Mechanical EngineeringDalian University of Technology
Awards and funding
Awards
- 2025
Best Paper AwardJournal of Smart Construction
- 2022
3rd Prize, Scan-to-BIM CompetitionIEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)
- 2021
2nd Prize, Scan-to-BIM CompetitionIEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)
Fellowships and grants
- 2024
UKRI Access to High-Performance Computing, principal investigatorProject: Construct highway digital twins from multi-modal data. Worth about 22,000 USD of GPU compute
- 2024
Sino-British Fellowship Trust Visiting FellowshipUniversity of Hong Kong
- 2023–2025
Marie Skłodowska-Curie Actions Future Road FellowshipUniversity of Cambridge
Talks and presentations
- 2026
Guest lecture, Digitization and AI in the Built EnvironmentCEE 121/221, Department of Civil and Environmental Engineering, Stanford University
- 2026
Paper presentationSASBE/FUTUREROADS/BDT Conference
- 2025
Showcase of the Road Research AllianceUK National Highways Development Centre
- 2025
RDFP programme workshop, two presentations; workshop hosted by DRF and RingwayUniversity of Cambridge
- 2024
Ordnance Survey Research Workshop; DRF Annual ReviewUniversity of Cambridge
- 2023
Ordnance Survey Research Workshop
- 2022
Paper presentationEuropean Conference on Computing in Construction (EC3), Rhodes, Greece
- 2021
Paper presentationEuropean Conference on Computing in Construction (EC3), online
- 2021
Presentation at NVIDIA GTCSession recording
Service
- 2025–present
Youth Editorial BoardJournal of Smart Construction
- 2019
Co-organizer, International Workshop on Built Environment Digital TwinningTUM Institute for Advanced Study and Siemens AG, Technical University of Munich