Derrick Mirindi

We present LoadTrace, a browser-based educational web application designed to improve how students learn gravity load takedown in multi-story building structures. The platform combines a parametric structural model, real-time visualization, stepwise explanation, and an artificial intelligence (AI)-supported tutoring agent in a single interface publicly deployed at https://derrickmirindi.github.io/loadtrace/. Rather than treating structural load analysis as a static sequence of hand calculations, the application reconfigures it as an interactive and interpretable learning process in which each floor's contribution to the total building load can be traced visually and numerically. We frame LoadTrace as a design-science artifact for structural engineering education rather than as a professional design package or a code-compliance engine. Its contribution lies in integrating three layers that are often separated in existing teaching tools: a transparent computational core, a narrative visual interface, and an AI-based conversational explainer grounded in the current model state. The application uses HTML (HyperText Markup Language), CSS (Cascading Style Sheet), JavaScript, and SVG (Scalable Vector Graphic) for the front-end implementation, while the conversational agent relies on Gemini-style API (Application Programming Interface) integration practices consistent with current developer guidance on API key handling and restriction. The visual structure of the interface also draws from grid-based editorial design principles informed by the Hyperagent public skill resource "skill-muller-brockmann-grid-systems.json", which was used as a design reference for layout discipline and typographic alignment. In addition, we argue that the significance of LoadTrace is not merely computational. Its broader value lies in demonstrating how AI evolution can reshape structural engineering education by making canonical load-path concepts more explorable, interpretable, and dialogic. In this sense, the application proposes a model for the next generation of structural learning environments: tools that do not replace engineering reasoning but amplify it through visual intelligence, interaction, and guided explanation.

Quantification and understanding of how heat, rainfall, and air quality vary within cities are needed to identify the area with the worst conditions, develop solutions to extreme weather, and assess the impact of proposed policies. However, neighborhood‐level variability is not well quantified because there are few environmental measurement stations within cities. In Baltimore City, a community‐based network of weather stations to address this issue has been developed through a partnership between universities, state agencies, and Baltimore residents. The weather stations are hosted by community partners, and the data collected are enabling the mapping of urban weather across the city and the testing of models and proposed mitigation strategies. In addition, the network provides direct community involvement, with resulting benefits of increased community engagement, education, and empowerment. Researchers have an opportunity to democratize the scientific process and ensure that local knowledge and lived experiences of city residents inform future decision‐making. The approach could be used as a model for other cities that apply similar monitoring instruments for other environmental exposures.