AI Infrastructure: Overview

Introduction

As we learned in the introductory lessons, creating smart applications means moving away from the traditional programming paradigm of hard-coded logic to generic algorithms that are hungry for data and computational power. This in turn means that we cannot use traditional architectures when designing AI-driven solutions; we need to introduce some new infrastructure blocks and tools.

First-timers are often surprised by how little time in a machine learning or AI project is spent actually doing model training or learning algorithms. As the diagram below depicts, a large part of developing AI applications means setting up the end-to-end pipeline and building blocks for enabling learning from data. While algorithms are at the core of the solution, they are brought to life by a multitude of techniques and tools around them.

Reality of AI projects: The hardest part of AI isn’t AI!