Kubernetes RAG Pipeline (MLOps)

A microservices-based Retrieval Augmented Generation (RAG) pipeline deployed on Kubernetes. This project demonstrates how to productionize AI models using Docker, Kubernetes, and Vector Databases.

🏗 Architecture

• Ingestion Service: Python FastAPI + Sentence Transformers.
• Vector Database: Qdrant (StatefulSet) for semantic search.
• Orchestration: Kubernetes (Minikube).

🚀 Key Features

• CPU-Optimized AI: Docker builds tuned for non-GPU inference.
• Resilience: Implemented livenessProbes and resource limits.
• Stateful Management: Persistent storage for vector data.

📦 How to Run

1. minikube start --memory=4096
2. kubectl apply -f k8s/
3. kubectl port-forward service/rag-embedding 7000:80