How LotteON built a personalized recommendation system using Amazon SageMaker and MLOps

This article discusses how LotteON built a personalized recommendation system using Amazon SageMaker and machine learning operations (MLOps).

Specifically, the article covers:

• Problem definition for building a recommendation system
• Solution architecture using AWS services like SageMaker, EMR, CodeBuild, S3, EventBridge, Lambda, and API Gateway
• Neural Collaborative Filtering (NCF) recommendation model
• MLOps components:
  • Data preprocessing with EMR
  • Automated model training and deployment with SageMaker Pipelines
  • Real-time inference with SageMaker endpoints
  • CI/CD structure integrating with GitLab and Jenkins
• Conclusion highlighting the benefits of using AWS services and SageMaker for MLOps