MLOps Starter Kit

A minimal hybrid Go/Python MLOps implementation demonstrating:

• Python model training with MLflow
• Go API gateway
• Kubernetes deployment
• Basic CI/CD workflow

Project Structure

Getting Started

chmod +x scripts/setup_mlflow.sh ./scripts/setup_mlflow.sh

Prerequisites

• Go 1.21+
• Python 3.9+
• Docker
• Kubernetes cluster (Minikube recommended)
• MLflow tracking server

Quick Start

1. Train model:

make train

2. Build and run Go API:

make build-go && ./bin/mlops-api

3. Test API: Test End-to-End

curl -X POST http://localhost:8080/predict \
  -H "Content-Type: application/json" \
  -d '{"features": [2.0, 3.0]}'

4. Deploy to Kubernetes:

make docker-build deploy

CI/CD Pipeline

GitHub Actions workflow included for:

# cd.yml
if: ${{ env.DEPLOY_ENABLED == 'true' }}

• Automated testing
• Docker image building
• Kubernetes deployment

Multi-Environment Support

strategy:
  matrix:
    environment: [staging, production]

• Development: Local development environment
• Staging: Kubernetes cluster with staging namespace
• Production: Kubernetes cluster with production namespace

Monitoring

Basic Prometheus/Grafana monitoring setup included in /deployments/monitoring

---

Verification Workflow

1. Train Model

# Start MLflow server
mlflow server --host 0.0.0.0 --port 5000

# In another terminal
make train

2. Test Go API

# Build and run Go API
make build-go
./bin/mlops-api

# Test endpoint
curl -X POST http://localhost:8080/predict \
  -H "Content-Type: application/json" \
  -d '{"features": [2.0, 3.0]}'

# Output: {"prediction": 5}

3. Deploy to Kubernetes

# Build Docker images
make docker-build

# Deploy
kubectl apply -f deployments/

# Check status
kubectl get pods

4. CI/CD Pipeline

# Automated testing
# Docker image building
# Kubernetes deployment

5. Monitoring

# Prometheus/Grafana monitoring

---

License

This project is licensed under the MIT License - see the LICENSE file for details.