MLflow Model Serving - Methods and Best Practices

Explore various methods of model serving with MLflow in this comprehensive one-hour video. Gain insights into both open-source MLflow and Databricks-managed MLflow approaches for serving models. Learn the fundamental differences between batch scoring and real-time scoring, with a particular focus on Databricks' upcoming production-ready model serving. Dive into topics such as prediction overview, vocabulary, online scoring, MLflow Model Server deployment options, container types, and scoring with JSON split-oriented format. Discover end-to-end ML pipeline examples, deployment plugins, and resources for MLflow Ray. Examine Keras/TensorFlow model formats and run model examples. Finally, explore model serving on Databricks, including production-grade model serving and the Databricks Model Serving launch.

Intro
Prediction overview
Vocabulary - Synonyms
Online Scoring
MLflow Model Server deployment options
ML.flow Model Server container types
Score with JSON split-oriented format
Python container
End-to-end ML Pipeline Example with MLflow
MLflow Deployment Plugins
MLflow Deployment Plugin Examples
MLflow Ray Resources
Keras/TensorFlow Model Formats
MLflow Keras/TensorFlow Run Models Example
Model Serving on Databricks
Databricks Production-grade Model Serving
Databricks Model Serving Launch