Home

Welcome to OpenRecSphere Wiki!

OpenRecSphere is a real-world, production-minded recommender system built with FastAPI, Redis/Faiss, PySpark, and deployed on a Kubernetes homelab cluster. This project covers everything from data pipelines to monitoring and A/B testing — designed to sharpen end-to-end ML engineering skills and showcase them publicly.

Architecture Diagram

[WIP] Coming soon

Technology Stack

Core

• FastAPI – high-performance Python API
• Redis / Faiss – fast similarity search
• PySpark – candidate generation / batch prep
• Docker – containerized microservices

Monitoring

• Prometheus – metrics collection
• Grafana – dashboards and observability
• Locust – load testing & traffic simulation

Deployment

• Docker Compose locally
• Prefect / Airflow – orchestration
• GitHub Actions – CI/CD for testing & linting