Developed an AI-driven spatial planning

Developed an AI-driven spatial planning tool that automates the creation of complete town master plans from a simple text brief. The system runs a local LLM (Llama 3.2:3B) fully on-device, ensuring data privacy with zero cloud dependency — ideal for government and municipal GIS workflows.

Key features built:

Natural language planning brief parser that converts requirements (population, area, facility counts, zoning restrictions) into a structured spatial plan Automated land-use allocation engine generating residential, commercial, industrial, institutional, and open-space zones on a configurable grid (e.g., 20m detail) Road and street network generation with automatic gap detection — flags disconnected street fragments and proposes link roads to connect isolated zones Facility placement logic for schools, clinics, police/fire stations, libraries, and places of worship, respecting environmental constraints Real-time validation report comparing generated output against a compliance ruleset, flagging unvalidated standards, area overruns, plot/dwelling mismatches, and unmet facility requirements Interactive map interface with togglable layers (land use, constraints, roads, blocks, plots, facilities, water/sewer mains) plus options to compare multiple town-shape scenarios and export planning files

Delivered as a desktop application combining GIS automation, generative AI.