Back to Projects Machine Learning

Real Estate Discovery Engine

Finding a home is about more than square footage and bedroom counts. It is about the neighborhood, the commute, the vibe. UrbanLiving wanted a search engine that felt human.

Map-Based Exploration

We abandoned the traditional list view in favor of a map-first approach. Users can sketch a shape on the map—say, a specific school district or a 10-minute walk from a subway station—and see only the homes inside that shape. This required heavy usage of Elasticsearch's geo-spatial querying capabilities, but the result feels effortless to the user.

Context is King

For each listing, we pull in data layers that matter:

  • Noise Levels: Is it near a busy highway?
  • Sunlight Analysis: Does the living room get afternoon sun?
  • Internet Speed: Is fiber available at this address?
Typical portals miss this. We made it front and center.

Technologies

Next.js Elasticsearch Google Maps API Firebase

Project Info

  • Sub-second search response times for over 2 million active listings.
  • "Commute Calculator" that sorts homes by travel time to the user's office.
  • Lead routing algorithm that matches buyers with the agent most experienced in that specific neighborhood.

Gallery