Quick Deploy Model Serving in Ranking Systems
Offered By: MLOps World: Machine Learning in Production via YouTube
Course Description
Overview
Explore a 32-minute conference talk from MLOps World: Machine Learning in Production on quick deployment of model serving in ranking systems. Discover how Yelp improved their process for updating and replacing models in ElasticSearch by developing an ES plugin integrated with their Model Platform. Learn about the implementation of Spark Pipelines trained and stored as MLeap bundles to MLFlow, allowing direct deployment to ES as MLeap pipelines. Understand how this approach encapsulates both the ML model and its feature engineering pipeline, enabling seamless swapping of modeling pipelines. Gain insights into the ES plugin's development and valuable lessons learned in optimizing model pipeline performance for efficient ranking systems.
Syllabus
Quick Deploy Model Serving in Ranking Systems
Taught by
MLOps World: Machine Learning in Production
Related Courses
Predicción del fraude bancario con autoML y PycaretCoursera Project Network via Coursera Clasificación de datos de Satélites con autoML y Pycaret
Coursera Project Network via Coursera Regresión (ML) en la vida real con PyCaret
Coursera Project Network via Coursera ML Pipelines on Google Cloud
Google Cloud via Coursera ML Pipelines on Google Cloud
Pluralsight