CAREER: Designing EdgeAI Tools With and For Communities in Low-Resource Settings

Across many rural and resource-constrained communities in the United States (including agricultural regions, Appalachian and coal country areas), resident-led groups and civic organizations work to address local concerns ranging from water safety and soil health to food production and land conditions. Yet gathering and acting on local data remains a persistent challenge. Existing tools for such tasks often assume reliable internet connectivity, use data platforms owned and managed by outside companies, and need specialized technical expertise that resource-constrained communities simply do not have. This leaves dedicated community groups without practical tools that work under real-world field conditions, regardless of how committed or knowledgeable they are about their local needs. This project develops a new class of low-cost, portable Edge Artificial Intelligence (EdgeAI) tools that embed machine learning directly into tiny, inexpensive computing devices that can operate without internet access or specialized equipment. These EdgeAI tools will enable grassroots communities with limited technical infrastructure to understand local conditions by creating, adapting, and managing their own data systems. Field research with civic organizations in Appalachian Tennessee will guide the design of these tools and generate design principles applicable to similar communities nationwide. By placing practical EdgeAI tools in the hands of community members, the project strengthens local capacity to detect and respond to concerns about issues such as water safety, soil health, and food production, supporting the health and economic welfare of communities that currently lack access to advanced technology. The project also engages undergraduate and K-12 students from rural and other low-infrastructure settings in hands-on, place-based computing, helping prepare the next generation of technologists who understand and serve the needs of their communities. This project investigates how participatory design methods and human-centered interfaces can enable non-expert users to deploy, adapt, and govern machine learning systems running on small, inexpensive embedded computing devices with tight memory and processing constraints (on the order of 256 kilobytes of memory). The research pursues four interconnected aims. First, participatory co-design methods will be developed and validated through field workshops, community mapping, and hands-on prototyping sessions where residents and civic organizations identify local data priorities and shape the design of EdgeAI tools for their specific contexts. Second, non-expert authoring interfaces will be created, allowing community members to collect and label data on-device, adjust detection thresholds, and retrain machine learning models directly on embedded devices without relying on desktop computers or internet connectivity. Third, these systems will be deployed at community gardens, water monitoring sites, and rural locations, where residents work with the tools under real field conditions over extended periods, revealing how community-led adaptation shapes system performance, trust, and local decision-making. Finally, education pathways will be built through undergraduate coursework and K-12 outreach, connecting students from rural areas and households without prior college experience to hands-on computing work rooted in the needs of their own communities. All tools, datasets, and curricula will be released as open-access resources to support adoption across a broad range of communities with limited technical infrastructure. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria. NSF Award ID: 2543328 | Program: 01002627DB NSF RESEARCH & RELATED ACTIVIT,01002930DB NSF RESEARCH & RELATED ACTIVIT,01003031DB NSF RESEARCH & RELATED ACTIVIT | Principal Investigator: Sai Swaminathan | Institution: University of Tennessee Knoxville, KNOXVILLE, TN | Award Amount: $376,586 View on NSF Award Search: https://www.nsf.gov/awardsearch/show-award/?AWD_ID=2543328 View on Research.gov: https://www.research.gov/awardapi-service/v1/awards/2543328.html