Deploying TensorFlow models with tfdeploy

Deploying TensorFlow models with tfdeploy

March 4, 2018

This talk presents tools and packages now available to the R community to test and deploy TensorFlow models at scale across services like: TensorFlow Serving, Google Cloud and RStudio Connect. This talks gives an overview on how to train a model in TensorFlow, Keras or TensorFlow Estimators, then explains how to export models with a common interface across all packages, covers testing the exported models locally and explains different deployment services available and use cases for each of them. This talk closes with a walkthrough in RStudio covering training, testing and deployment. It also briefly covers an additional deployment alternative using kerasjs and answers a few questions from the audience.

View Slides

About the speaker

Javier Luraschi

Javier is experienced in technologies ranging from desktop, web, mobile and backend; to augmented reality and deep learning applications. He previously worked in Microsoft Research and SAP and holds a double degree in Mathematics and Software Engineering. Javier is the creator of packages like sparklyr, r2d3, cloudml and author of "Mastering Spark with R".