QuickFind

The Problem

Finding specific items—like a “2-bedroom apartment” or a “used iPhone”—in the current digital landscape often requires manually scrolling through dozens of disorganized, spam-filled social media groups. These platforms lack structured data, making accurate searching and filtering nearly impossible.

The Solution: AI-Powered Structuring

QuickFind solves this by treating social media as a structured database. It doesn’t just scrape data; it understands it. By leveraging Google’s Gemini 2.0 Flash LLM, the system reads raw, unstructured post text (e.g., “pwleitai iphone, timi inbox”) and intelligently extracts the intent, normalizes the price, and maps the location.

System Architecture & Tech Stack

The platform is built on a microservices-lite architecture, orchestrated via Docker Compose on a Hetzner VPS, separating the user-facing application from the heavy-lifting backend workers.

• Frontend (React/Vite): A modern, responsive UI with dark mode, smart filters, and real-time alerts.
• Backend & Database (Supabase/PostgreSQL): Handles authentication, real-time webhooks, and complex relational data storage.
• The AI Pipeline (Python 3.11):
  • The Orchestrator & Scraper: Uses undetected-chromedriver for headless browsing and smart deduplication to bypass bot detection and ingest raw data.
  • The AI Processor: Batches raw data and sends it to the Gemini API to categorize, format, and structure the listings.
  • The Validator: A self-healing script that continuously checks active listings to mark them as ‘Sold’ or ‘Deleted’ and permanently archives expiring CDN images to a self-hosted SFTP server.

Key Features for Users

• Unified Feed: Aggregates listings from hundreds of disparate groups into a single, clean interface.
• Real-Time “Agents”: Users set specific criteria (e.g., Budget: 500€, Location: Athens), and the system monitors the database 24/7, sending instant alerts when a match is found.
• Permanent Image Hosting: Solves the issue of expiring social media image links through an automated archiving process.

The Result

QuickFind successfully transforms chaotic, unstructured social feeds into a clean, searchable, and highly efficient marketplace aggregator, saving users countless hours of manual searching.