Project
An AI-powered rental property aggregator that scrapes listings from multiple sources, scores photo quality with Gemini Vision, and presents everything on an interactive map-and-table dashboard.
How it works
Crawls Craigslist, Zillow, Apartments.com, Trulia, Realtor.com, and Zumper. Extracts price, beds, baths, sqft, amenities, photos, and contact info.
Each listing's photos are analyzed by Google's Gemini Vision API for quality rating, flooring type, natural light, kitchen style, outdoor space, and size impression.
Sortable table with filters for price, distance, beds, quality, flooring, amenities. Map with clustered markers, photo popups, and housing-type color coding.
Add work locations and see average commute distance for every listing. Prioritize apartments that minimize your daily drive.
Flags suspicious listings based on price anomalies, stock photos, and known scam patterns from the data.
Every listing gets a generated detail page with full photo gallery, lightbox, formatted description, contact info, and commute breakdown.
Current state
This one was built to solve an actual apartment hunt, and it did. The live demo is the June 2026 snapshot of that search: around 450 generated detail pages over the units it had found, with the map, the filters, the photo scores and the commute breakdowns all working against real scraped data. The scrapers still run, but nothing new is being crawled into it, so treat the listings as a preserved dataset rather than a current market. The interesting part was never the listings anyway; it was watching a vision model read a kitchen photo and be right about it.
Built with