Developing Workforce-Informed Digital Twins for Smart Redevelopment Site Classification
PI: Yuqing Hu (Architectural Engineering)
Funding tuition will be covered through PI’s start-up.
Team Members and Mentors: Dr. Miller Scarlett, Industrial and Manufacturing Engineering (Mentor)
Project Description
Post-industrial regions in the U.S.—particularly in Appalachia—face long-standing challenges stemming from the decline of coal, steel, and manufacturing industries. In response, national initiatives such as the DOE Energy Communities Initiative, EPA Brownfields Program, and the broader push to reshore domestic manufacturing are driving significant interest and investment toward repurposing underutilized industrial sites. However, many municipalities and regional planning agencies lack data-driven tools to evaluate which sites are best suited for redevelopment, particularly with respect to workforce access, mobility infrastructure, proximity to essential services, and alignment with training facilities. This project addresses that gap by developing a graph-based digital twin framework to classify and prioritize redevelopment sites based on workforce and infrastructure readiness. The digital twins will integrate real-world mobility data, multimodal transportation networks, service infrastructure, and local workforce training assets to simulate and evaluate redevelopment potential. A key technical innovation is the use of graph similarity modeling, which compares candidate sites to a curated set of previously successful redevelopment examples, identifying key structural and spatial features that contribute to long-term workforce viability. The goal of this ICDS-supported phase is to prototype the digital twin system, implement and test similarity metrics, and generate preliminary modeling results across a small set of pilot sites. These outputs will support a future NSF Smart and Connected Communities proposal. To achieve this objective, key technical tasks for the ICDS Junior Researcher include:
• Aggregating and preprocessing geospatial datasets (SafeGraph mobility, OpenStreetMap and GTFS transit networks, USPS vacancy, ACS demographics, and workforce training facility data)
• Constructing multi-layer digital twin graphs where nodes represent services, job centers, training facilities, and transit stops, and edges capture multimodal travel time or service gravity
• Implementing graph neural network (GNN) embeddings (e.g., node2vec, GraphSAGE) and graph similarity functions (e.g., Weisfeiler-Lehman kernel)
Desired Skills and Expertise
• Programming experience in Python, particularly with libraries such as PyTorch Geometric, DGL, or NetworkX; Familiarity with spatial data analysis, geospatial APIs, or urban infrastructure modeling
• Experience or interest in graph neural networks (GNNs) or graph-based similarity methods • Comfort working with open geospatial datasets (e.g., OSM, GTFS, ACS) and mobility data Other Requirements
• Preferred: Post-comps PhD student, such as Architectural Engineering, CS, or IST.
Objectives of Work Supported by This Call
Prepare preliminary work for NSF proposals, infrastructure system and people program, or smart and connected communities’ program (Due Jan 12, 2026)
Medium- to Long-Term Goal
To establish a scalable, AI-enabled decision-support framework that enables planners to assess and prioritize redevelopment sites based on workforce accessibility and infrastructure readiness.