TensorFlow on AWS

Shashank Prasanna (Amazon Web Services), vikrant kahlir (Amazon Web Services), Rama Thamman (Amazon Web Services), Shreyas Subramanian (Amazon)

9:00am—5:00pm Monday, October 28—Tuesday, October 29

Location: Room 211

Average rating:

(3.00, 2 ratings)

Participants should plan to attend both days of training course. Note: to attend training courses, you must be registered for a Platinum or Training pass; does not include access to tutorials on Tuesday.

Amazon Web Services (AWS) offers a breadth and depth of services to easily build, train, and deploy TensorFlow models. Shashank Prasanna, Vikrant Kahlir, and Rama Thamman give you hands-on experience working with these services.

What you'll learn, and how you can apply it

Discover how to easily build, train, and deploy TensorFlow models on AWS

Who is this presentation for?

You're a data scientist, ML developer, or researcher.

Level

Intermediate

Prerequisites:

A basic understanding of TensorFlow

Hardware and/or installation requirements:

A laptop
An AWS account

Outline

Day 1

Build, train, and deploy TensorFlow models with Amazon SageMaker
A fully managed machine learning platform, Amazon SageMaker enables developers and data scientists to build, train, and deploy custom TensorFlow models. You’ll learn how to build custom TensorFlow models in Jupyter and iterate fast in Amazon SageMaker’s local mode, pass TensorFlow scripts to Amazon SageMaker to train your model on a cluster of instances, and deploy your trained model to Amazon SageMaker hosting and seamlessly specify preprocessing and postprocessing steps.

Reduce inference cost by up to 75% for TensorFlow models with Amazon Elastic Inference
You’ll learn how to deploy your TensorFlow 2.0 models with Amazon Elastic Inference, allowing you to attach just the right amount of GPU-powered acceleration to any Amazon SageMaker or Amazon EC2 instance and set up Elastic Inference accelerators for your TensorFlow model and measure your cost savings and performance.

Day 2

Distributed training with TensorFlow models on Amazon SageMaker and Amazon Elastic Container Service for Kubernetes (EKS)
Reducing time to train your TensorFlow models is crucial in improving your productivity and reducing your time to market. You’ll learn how to efficiently scale your training workloads to multiple instances, with Amazon SageMaker doing the heavy lifting for you, and how to set up and optimize distributed training on a Kubernetes cluster in Amazon EKS with AWS Deep Learning Containers.

Train once and deploy everywhere on cloud and edge devices with SageMaker Neo
You’ll learn how to use Amazon SageMaker Neo Deep Learning Compiler (DLC) to compile your trained TensorFlow models and deploy them in the cloud or on edge devices using AWS IoT Greengrass, how Neo DLC optimizes the trained models by improving efficiency and reducing memory footprint of the compiled model, how Neo runtime abstracts the underlying hardware and allows running the compiled model on multiple hardware devices such as Intel Xeon and Atom, NVIDIA Jetson, ARM, and many more, and you’ll gain experience in improving runtime performance by 2x and reducing memory footprint by 10x using SageMaker Neo.

About your instructors

Shashank Prasanna is a senior AI and machine learning evangelist at Amazon Web Services, where he focuses on helping engineers, developers, and data scientists solve challenging problems with machine learning. Previously, he worked at NVIDIA, MathWorks (makers of MATLAB), and Oracle in product marketing and software development roles focused on machine learning products. Shashank holds an MS in electrical engineering from Arizona State University.

Vikrant Kahlir is a solution architect at Amazon Web Services.

Rama Thamman is an R&D manager at Amazon Web Services.