TaxSale Application
Description: A full-stack web application for discovering and managing tax sale properties across multiple Georgia counties. Parses county PDF and CSV listings, geocodes addresses onto an interactive map, and provides filtering, search, and admin tools for managing data sources.
Role: Full-Stack Developer
Technologies:
Node.js
Express
SQLite
Python
Leaflet.js
PDF.js
OpenStreetMap
bcrypt
Features
- Interactive map with geocoded tax sale properties using Leaflet.js and OpenStreetMap
- Multi-county support with automated PDF and CSV parsing (Chatham County, DeKalb County)
- Property search and filtering by county, zip code, address, and tax amount
- Admin dashboard with parser management, database tools, and scheduled data refresh
- User authentication system with registration, login, and role-based access
- Google Street View integration for property previews
Architecture
- Frontend: Vanilla JavaScript with Leaflet.js for mapping and PDF.js for document rendering
- Backend: Node.js/Express server with RESTful API endpoints and session-based authentication
- Database: SQLite for property data, user accounts, and county configurations
- Data Pipeline: Python-based PDF parsing (PyPDF2, pdfplumber) with Node.js CSV parsing fallback
- Security: Helmet.js, CORS, rate limiting, bcrypt password hashing, and input validation with Joi
Challenges
- Parsing inconsistent PDF formats across different county tax offices required a factory pattern with multiple parser strategies
- Geocoding hundreds of addresses reliably while respecting API rate limits from OpenStreetMap
- Setting up a dual-language environment (Node.js + Python) for deployment on Render
- Designing a scalable county registration system to support adding new counties without code changes
Outcomes
- Successfully parses and maps 200+ tax sale properties across multiple counties
- Automated data refresh pipeline eliminates manual data entry
- Admin interface enables non-technical users to manage county data sources and trigger refreshes
- Extensible architecture allows new counties to be added through configuration rather than code
Lessons Learned
- Building resilient parsers with graceful fallback mechanisms is essential when dealing with inconsistent external data sources
- The factory pattern proved invaluable for handling multiple data formats from different counties
- SQLite is an excellent choice for single-server applications — simple deployment with no external database dependency
- Designing admin tools early saves significant time during development and testing