Warsaw Real Estate Price Predictor (REST API & LLM Observability)
2026-09-14
An end-to-end production-grade REST API designed to estimate property prices in Warsaw. The system combines traditional machine learning for valuation with OpenAI GPT-4o-mini for structured feature extraction from raw Polish listings, fully monitored via Langfuse telemetry.
🔗 Quick Links & Live Demo
- Interactive API Documentation (Swagger UI): https://warsaw-real-estate-api.onrender.com/docs
- GitHub Repository: https://github.com/Vados182/Warsaw-real-estate-api.git
🏗️ Technical Architecture
The application is built around a modern microservice architecture tailored for production reliability:
- FastAPI Framework: Exposes
/predict(tabular input) and/predict-from-text(unstructured listing description) endpoints. - LLM Feature Extraction (OpenAI + Instructor): Uses
GPT-4o-miniwrapped with theInstructorlibrary andPydanticschemas to deterministically parse raw text descriptions into typed parameters (district,area_sqm,rooms,floor). - Machine Learning Pipeline (PyCaret): Predicts final property valuation using a tuned regression model trained on historical Warsaw real estate data.
- Observability & Monitoring (Langfuse): Real-time tracking of API latency, token consumption, error rates, and costs per request.
- Deployment: Fully containerized with Docker and hosted on Render.com.
🚀 Key Features & Highlights
- Deterministic JSON Output: Guarantees strict type validation on LLM responses to prevent runtime errors in the ML pipeline.
- Production Observability: Full visibility into LLM calls, tracking costs, and system performance via Langfuse.
- Auto-Generated Docs: Interactive Swagger UI documentation available out of the box for testing endpoints.
💻 Tech Stack
- Core: Python 3.11, FastAPI, Uvicorn, Pydantic
- Machine Learning & AI: PyCaret, OpenAI (GPT-4o-mini), Instructor
- Observability: Langfuse
- DevOps & Hosting: Docker, Render, Git