The AI Toolkit for IBM Z and LinuxONE is a comprehensive suite of tools designed to streamline the development and deployment of AI models on IBM's enterprise-grade platforms. This toolkit empowers developers and data scientists to leverage the power of IBM Z and LinuxONE systems for AI workloads, offering a seamless integration with popular AI frameworks and libraries. It includes features such as automated model deployment, performance optimization, and security enhancements, making it an ideal choice for organizations seeking to harness the potential of AI in their enterprise environments.

In the rapidly evolving landscape of technology, the ability to harness the power of artificial intelligence and machine learning has become paramount. This IBM Redbooks publication provides a guide to leveraging IBM Z systems for accelerated AI workloads, offering insights into various tools and frameworks that can significantly enhance performance and efficiency. As we delve into the intricacies of fraud detection and the application of deep learning models, readers will gain a deeper understanding of how IBM Z can be a game-changer in this domain.

In this IBM Redbooks publication, we discuss the architecture, components, and deployment strategies of the AI Toolkit for IBM Z and LinuxONE. We explore how this toolkit integrates with open-source frameworks such as TensorFlow, PyTorch, Snap ML, and ONNX, and how it leverages IBM Z hardware accelerators like the Telum and Spyre processors to deliver high-performance AI inferencing.

This publication also provides practical guidance through real-world use cases, including near real-time credit card fraud detection, to demonstrate how AI models can be trained, optimized, and deployed efficiently on IBM Z. Readers will gain insights into model conversion, quantization, performance tuning, and containerized deployment using IBM’s customized backends for Triton Inference Server and TensorFlow Serving.

Whether you are a data scientist, AI engineer, system architect, or IT decision-maker, this Redpaper will help you understand how to build and scale AI solutions on IBM Z and LinuxONE platforms with confidence, security, and enterprise-grade support.

Chapter 1. Introduction to AI on IBM Z

Chapter 2. Overview of fraud detection use case

Chapter 3. Introduction to PyTorch

Chapter 4. IBM Z Deep Learning Compiler

Chapter 5. IBM Snap ML

Chapter 6. IBM Z Accelerated for NVIDIA Triton Inference Server

Chapter 7. IBM Z Accelerated for TensorFlow

Chapter 8. IBM Z Accelerated Serving for TensorFlow

Chapter 9. Other considerations