Legible Landscapes
Global biodiversity and environmental data are abundant and free, but generating insights from them often requires data science and GIS expertise that most decision-makers, conservation practitioners, and community members don’t have. The gap is usually not in data availability, but in legibility.
My work in this area focuses on building web tools that lower the technical barrier between people and biodeversity data. Some projects use LLMs to enable plain-language queries, while others take a more traditional dashboard approach. What ties them together is a commitment to meeting users where they are and designing around the questions they actually want to ask.
Current projects include:
- An LLM-augmented web tool that lets urban planners and city agencies ask natural-language questions of biodiversity and environmental data across sources like GBIF, OpenStreetMap, and US Census data. This project uses open-weight language models and a custom Model Context Protocol (MCP) server, and is being piloted with the Reimagining SF coalition.
- A fire followers dashboard built in collaboration with the California Native Plant Society, which turns data from their Fire Followers iNaturalist campaign into an interactive exploratory tool.