Photo of Qianchen Yu Research interests

Reinforcement learning–driven resource allocation in complex urban systems

Ph.D. student in Urban and Regional Planning at the University of Florida — working on AI-driven simulation and reinforcement learning for post-disaster housing recovery, urban resilience, and construction scheduling.

PyRebuild simulator RL for reconstruction scheduling GeoAI & discrete-event simulation

Research

PyRebuild: Simulation and Deep RL for Post-Disaster Housing Reconstruction

PyRebuild is a Python-based discrete-event simulator that models post-earthquake housing reconstruction under resource constraints and uncertainty. It serves as an AI-ready testbed for studying how different scheduling policies affect system-wide recovery, regional disparities, and household-level fairness.

  • Models batched damage reporting, dynamic crew mobilization, and lognormal repair durations calibrated to real events (e.g., the 2018 Lombok earthquake).
  • Trains PPO-based reinforcement learning policies (Stable-Baselines3) for crew allocation and compares them to rule-based baselines (LJF, SJF, random).
  • Encodes fairness through waiting-time-based rewards and candidate ordering, enabling analysis of efficiency–equity trade-offs.
  • Explores transfer-learning setups where policies learned from one region are adapted to regions with different damage structures and resource levels.

Academics

Ph.D. in Urban and Regional Planning
University of Florida, College of Design, Construction and Planning · 2024.08 – Present
Gainesville, Florida, USA
  • Research on post-disaster housing recovery, urban resilience, and RL-driven resource allocation.
  • Developing the PyRebuild simulator and PPO-based policies for reconstruction scheduling.
Master of Regional Planning
Cornell University · 2022.08 – 2024.05
Ithaca, New York, USA
  • Concentration in urban data science and spatial analysis, with strong emphasis on Python-based spatial methods and GeoAI.
B.Eng & B.A in Urban Planning and Design
Xi'an Jiaotong-Liverpool University (joint program with University of Liverpool) · 2018.09 – 2022.07
Suzhou, China / Liverpool, UK
  • Training in urban economics, social inclusion, real-estate analysis, and multi-scale planning and design.

Conferences & Workshops

CRC&CI 2026 (ASCE)

Deep RL-based Transferable Scheduling for Post-Disaster Housing Reconstruction

Oral · Construction Research Congress & Computing in Civil Engineering · San Antonio, USA · March 2026 (forthcoming)

  • Extends PyRebuild with deep RL under stochastic disruptions and explores cross-region policy transfer.

ACSP 2025

Adaptive Post-Disaster Housing Reconstruction Scheduling: A Deep Reinforcement Learning Approach

Oral · Association of Collegiate Schools of Planning Annual Conference · Minneapolis, USA · October 2025

  • Discusses adaptive reconstruction schedules under changing resource and policy scenarios from a planning perspective.

i3CE 2025 (ASCE)

PyRebuild: A Simulation Framework for AI-Driven Adaptive Resource Allocation in Post-Disaster Housing Recovery

Talk · International Conference on Computing in Civil Engineering · New Orleans, USA · May 2025

  • Introduces the PyRebuild framework and evaluates rule-based scheduling strategies under incomplete information.

AAAI 2025 AI4UP Workshop

PyRebuild: A Python-Based Simulator For Dynamic Post-Earthquake Reconstruction

Poster · AI for Urban Planning (AI4UP) · Philadelphia, USA · March 2025

  • Presents PyRebuild as an AI-ready testbed for RL-based scheduling and fairness-aware resource allocation.

Experience

Geospatial Analyst
uTECH Lab, Cornell University · 2023.09 – 2024.01
Ithaca, New York, USA
  • Contributed to a geospatial visualization framework for traffic–human–environment studies.
  • Built multi-source spatial datasets linking socio-economic indicators with road networks and land-use layers.
Data & Geospatial Analyst Intern
CityDNA · 2022.06 – 2024.08
Beijing, China
  • Assisted in building urban spatial databases for site analysis and urban renewal projects.
  • Developed mapping workflows to turn analytical results into clear maps and graphics for planning practice.
Research Assistant
Xi'an Jiaotong-Liverpool University · 2021.01 – 2021.04
Suzhou, China
  • Collected housing transaction and income data for eight major Chinese cities.
  • Constructed a housing affordability database and performed basic descriptive analysis and visualization.

Skills

Programming & ML / RL

  • Python (NumPy, pandas, PyTorch).
  • Stable-Baselines3 (PPO, DQN) for RL training on custom simulation environments.
  • R for statistical modeling and regression analysis.
  • Git-based workflows for experiment tracking and reproducible pipelines.

Data & Spatial Analysis

  • Data cleaning, descriptive statistics, basic modeling, and visualization.
  • ArcGIS Pro and QGIS for spatial data processing, mapping, and urban analysis.
  • Integration of socio-economic indicators with spatial units for planning and policy questions.

Research & Communication

  • Experience writing research briefs, academic-style reports, and slides in English and Chinese.
  • Comfortable explaining technical results to planning, policy, and interdisciplinary audiences.

Publications

Yu, Q., Alisjahbana, I., & Wong, V. W. H. (2025). PyRebuild: A Simulation Framework for AI-Driven Adaptive Resource Allocation in Post-Disaster Housing Recovery. In Proceedings of the ASCE International Conference on Computing in Civil Engineering (i3CE 2025), New Orleans, USA.

📄 PyRebuild (i3CE 2025)

Yu, Q., Alisjahbana, I., & Wong, V. W. H. (2026). Optimizing Post-Disaster Housing Reconstruction Scheduling under Stochastic Disruptions using Deep Reinforcement Learning. Accepted for publication in the ASCE Construction Research Congress & Computing in Civil Engineering (CRC&CI 2026) conference proceedings, San Antonio, USA.

Full text coming soon (camera-ready in preparation).

Let’s connect

I am happy to discuss collaborations or internships at the intersection of urban resilience, AI-driven simulation, reinforcement learning, and social/urban computing.

📧 Email LinkedIn ↑ Back to top