Cloud-Native Emotion ML Inference Service

This project is a containerized Machine Learning Microservice deployed on Kubernetes. It exposes an API for emotion recognition (predicting emotions from input data) and is backed by a MongoDB database for persistence and state management.

Architecture Overview

The infrastructure is defined using Kubernetes manifests and consists of the following components:

1. Emotion API (Deployment & Service)

   • Image: shiva2020/emoimage:latest
   • Description: The core machine learning inference service running as a stateless REST API.
   • Internal Routing: Exposes port 8080 internally and is mapped to port 80 by its ClusterIP Service.

2. MongoDB Database (Deployment & Service)

   • Image: mongo:latest
   • Description: A standard MongoDB instance for storing API metadata, prediction logs, or user records.
   • Security: Secured via Kubernetes Secrets mapped as environment variables.

3. Ingress

   • Host: emotion-api.local
   • Description: Routes external HTTP traffic to the internal emotion-api ClusterIP service.

Prerequisites

• A running Kubernetes cluster (Minikube, Docker Desktop, EKS, GKE, etc.)
• kubectl command-line tool configured to communicate with your cluster.
• An Ingress Controller (e.g., NGINX Ingress Controller) installed on your cluster.

Deployment Instructions

1. Configure Local DNS (Optional) If you are running this locally, map the Ingress host to your localhost (or Minikube IP) in your /etc/hosts file:

   127.0.0.1 emotion-api.local

2. Deploy the configuration Apply the Kubernetes manifests in the following order to ensure dependencies (like secrets) exist before the deployments are created:

   # 1. Apply Secrets
   kubectl apply -f k8s/secret.yaml
   
   # 2. Deploy the Database
   kubectl apply -f k8s/mongodb.yaml
   
   # 3. Deploy the Machine Learning API
   kubectl apply -f k8s/deployment.yaml
   kubectl apply -f k8s/service.yaml
   
   # 4. Configure External Access
   kubectl apply -f k8s/ingress.yaml

3. Verify the deployment Ensure all pods are in the Running state:

   kubectl get pods,svc,ingress