This is a full-stack application built for the Starboard AI Take-Home Challenge. It demonstrates comparable analysis for industrial properties using Cook County data.
To fetch and display comparable industrial property records based on user queries. Our primary goal was to simulate an agent-based system capable of providing real estate investors with intelligent property comparables.
- Next.js (App Router)
- React 19 (Client Components)
- Tailwind CSS
- Node.js + API Routes
- CSV File as Mock API Source
- Deployment: Vercel
- User Input: User enters a location, zip code, or keyword (e.g.,
Cicero,warehouse). - Backend Logic: Filters data using fuzzy match and weights fields like
zip,building area, andzoning. - Comparable Generation: Generates a confidence score based on similarity in key fields.
- Results Display: Lists top comparables in descending order of match score.
We explored integrating real-time APIs across county sources and commercial platforms, including:
- Cook County Open Data:
datacatalog.cookcountyil.gov - Realtor.com and Zillow (via headless scraping)
- ATTOM Data API
- LoopNet (headless scraping via Puppeteer and Playwright)
| Issue | Description |
|---|---|
| ❌ Strict Rate Limits | Most APIs limited throughput or required whitelisting. |
| ❌ Authentication Hurdles | Many APIs required application approval for keys. |
| ❌ Schema Inconsistencies | Field names varied (sqft, area, bldg_area), requiring mapping logic. |
| ❌ Blocked Bots | Realtor, LoopNet blocked Puppeteer/Playwright scraping despite stealth efforts. |
| ❌ Data Gaps | Zoning codes and industrial classifications were often missing or unreliable. |
💡 Solution: We pivoted to using a cleaned CSV dataset scraped from Cook County and simulated the agent logic over this structured offline data.
- File:
clean_industrial_properties.csv - County: Cook County, IL
- Size: ~400 records
- Columns:
address,zip,building_area,zoning_code,property_class, etc. - Preprocessed and cleaned using Python to ensure consistent types and remove invalid entries.
🎥 Click to Watch the Loom Demo
🔗 Live Deployment on Vercel
git clone https://github.com/rushi-k12/Starboard-AI.git
cd Starboard-AI
npm install
npm run dev