About Zynki: Real Estate AI Platform

Zynki is an AI-powered real estate intelligence platform developed as a final project for an academic course. The platform leverages advanced machine learning techniques to predict Israeli real estate prices with high accuracy.

Our system uses an ensemble model combining Random Forest and XGBoost algorithms, trained on real Israeli market data from Nadlan.gov.il (Israel Land Authority). The ensemble approach achieves 95%+ accuracy (R² 0.9524) with a Mean Absolute Error of ₪315,815.

The platform is built using a CrewAI Flow architecture with 2 specialized crews:

• Data Analyst Crew: Handles data collection, cleaning, and exploratory analysis
• Data Scientist Crew: Performs feature engineering, model training, and evaluation

The technology stack includes React 18 (PWA), FastAPI, CrewAI 1.9.3, and modern ML libraries (scikit-learn, XGBoost). The platform provides instant price predictions, confidence scoring, and market insights for properties across Israeli cities.