Infrascope AI

Inspiration

Inspiration

Urban planners and decision-makers often lack real-time, AI-powered tools to determine the best land use or infrastructure type for a given area. InfraScope AI solves this by analyzing satellite imagery, terrain, and accessibility data to recommend optimal infrastructure placements.

What it does

InfraScope AI allows users to draw an area on a map and instantly receive recommendations on what infrastructure (school, hospital, green space, etc.) would be most suitable based on terrain, existing facilities, population data, and land use.

How we built it

• Frontend: Built with React and Mapbox for location selection and visualization.
• Backend: FastAPI serves as the API layer for AI model requests.
• AI: Google Cloud Vertex AI processes satellite data and predicts optimal land use.
• Database: MongoDB stores infrastructure data with geospatial indexing.
• CI/CD: GitHub & Cloud Build for continuous deployment.
• Deployed on: Google Cloud Run.

Challenges we ran into

• Integrating various geospatial data sources and cleaning them quickly.
• Optimizing AI models to produce useful infrastructure suggestions.
• Deploying a dual frontend-backend setup with Docker on Google Cloud.

Accomplishments we’re proud of

• Real-time geospatial ML working end-to-end
• Live frontend map interface
• Cloud-native deployment under 24 hours

What we learned

• Working with geospatial ML using Google Cloud Vertex AI
• MongoDB geospatial querying and vector search
• End-to-end deployment using Docker and Cloud Run

What’s next

• Expand to more cities and datasets
• Enable collaborative planning sessions
• Add PDF export and growth forecast features es

How we built it

Challenges we ran into

Accomplishments that we're proud of

What we learned

What's next for Infrascope AI

Built With

• ai
• docker
• fastapi
• gcp
• gemini
• github
• mapbox
• mongodb
• react
• vertex