Machine Learning Model Deployment for Climate Research

Create a serverless Node.js microservice architecture for deploying and managing machine learning climate prediction models using TensorFlow.js. Design a system that can dynamically load pre-trained models, support model versioning, provide automated performance monitoring, and generate comprehensive inference reports. Include robust error handling for various climate dataset formats and implement secure model serving with authentication.