hops-online-inference

What problem does it solve?

This Skill allows for deploying models and conducting online inference directly in the Hopsworks environment, streamlining the process of real-time model deployment and prediction.

Core Features & Use Cases

• Real-time Model Deployment: Deploy models and create running KServe endpoints for immediate inference.
• Feature Store Integration: Utilize the Hopsworks Feature Store for online feature lookups during inference.
• Predictor Scripting: Execute custom Python scripts to extend functionality and control model prediction logic.
• Use Case: When building a recommendation engine for e-commerce, deploy a trained model using this Skill to provide real-time product recommendations to users.

Quick Start

Deploy a trained model and make a prediction by executing: model.deploy('recommendation_service', script_file='predictor.py')