CITY-AI Education Modules
The CITY-AI curriculum aims to bridge the gap between urban science, cyberinfrastructure (CI), and artificial intelligence (AI) using the HPC resources including I-GUIDE Platform and NSF Anvil Supercomputer.
Curriculum Overview
All training materials are developed as Open Educational Resources (OERs) following FAIR (Findability, Accessibility, Interoperability, and Reusability) principles, packaged with interactive Jupyter Notebooks and containerized Docker environments.
Theme 1: Core CI Skills
Linux, HPC, Python, Profiling & Scaling, NSF Anvil Supercomputer
Theme 2: Urban Analytics
Spatial Data, Big Data, Streaming IoT, Spatiotemporal Analysis
Theme 3: Urban Use-Cases
Air Quality, Urban Heat Islands, GeoAI, SVI & Computer Vision
Theme 1: Core Cyberinfrastructure (CI) Skills
These modules lower technical barriers for urban planners and geographers by introducing high-performance computing (HPC) environments, terminal workflows, and distributed resources.
| Module Topic | Level | Key Learning Goals |
|---|---|---|
| Intro to Linux | Beginner | Terminal navigation, shell scripting, file management, I/O streams, and cluster environment basics. |
| Intro to HPC | Beginner | Accessing NSF ACCESS resources, NSF Anvil supercomputer, OpenOnDemand portal, and batch job scheduling (SLURM). |
| Python & JupyterLab | Intermediate | Python programming in JupyterLab; scientific libraries including NumPy, SciPy, Pandas, and Matplotlib. |
| Profiling & Scaling | Advanced | Computational complexity analysis, estimating resource requirements, parallelization, and software scaling strategies. |
Theme 2: Urban Data Analytics
Focused on building fundamental geospatial, spatiotemporal, and high-velocity streaming data processing skills.
| Module Topic | Level | Key Learning Goals |
|---|---|---|
| Spatial Data | Beginner | Core GIS concepts, spatial reference systems, vector/raster data structures, and interactive thematic mapping. |
| Big Data | Intermediate | Handling high-volume urban datasets, memory-efficient processing, spatial indexing, and distributed storage. |
| Streaming Data | Intermediate | Acquiring, cleaning, and ingesting real-time urban data streams from sensor networks and IoT gateways. |
| Spatiotemporal (ST) Data | Advanced | ST analysis techniques, clustering algorithms, spatiotemporal autocorrelation, and dynamic geovisualization. |
Theme 3: Urban Informatics Use-Cases
Hands-on case studies demonstrating how cutting-edge GeoAI and CI solve urgent smart city and environmental challenges.
| Module Topic | Level | Key Learning Goals |
|---|---|---|
| Volunteered Geographic Info (VGI) | Intermediate | Extracting, cleaning, and modeling crowdsourced geospatial data from OpenStreetMap (OSM) for urban infrastructure. |
| Air Quality Monitoring | Intermediate | Processing spatio-temporal streams from environmental sensors and predicting localized air pollution exposure. |
| Urban Heat Island (UHI) | Advanced | Integrating satellite thermal observations with in-situ urban sensor networks to model microclimate temperature anomalies. |
| Geospatial AI (GeoAI) | Intermediate | Formulating machine learning models that explicitly account for spatial heterogeneity and temporal dynamics. |
| Transportation & Accessibility | Advanced | Utilizing real-time GTFS feeds to analyze public transit delays and compute scalable spatial accessibility metrics. |
| Computer Vision (CV) | Advanced | Applying deep learning vision architectures (CNNs, Vision Transformers) to fine-tune image models on urban datasets. |
| Street View Imagery (SVI) | Advanced | Automated object detection and segmentation on street-level panoramas to identify urban physical retrofitting and greenery. |
| Urban Remote Sensing | Advanced | High-resolution satellite and LiDAR image processing for urban land use classification and 3D canopy extraction. |
Computing Environment
All training modules are hosted and executable directly through:
- I-GUIDE Platform & JupyterHub: Providing reproducible, containerized cloud environments with pre-installed GeoAI libraries.
- NSF Anvil Supercomputer (ACCESS): Accelerating intensive training with large-capacity GPU nodes (NVIDIA A100/H100) and Kubernetes-based composable services.