ML and AI are undoubtedly becoming increasingly important. However, integrating ML/AI into software is not an easy task. Most projects in this context are prototyped in Python. However, a real solution has numerous requirements that go beyond the scope of an experimental application. ML.NET is a viable approach that also has its limits.
In this talk, you will learn how to implement real projects using Microsoft Machine Learning .NET frameworks. And not only that. We will show how a proper architecture of such a solution is created step by step. You will first learn how ML.NET works. Next, we show how "heavy workloads" such as ML training process can be implemented why technologies such as Docker can sometimes be surprisingly unhelpful here and how such "limits" can be circumvented using Azure Batch Service. That's not all. The prediction part of such an application can also make your web-hosted web application go haywire.
This is an advanced talk with lots of demos and insights into the world of ML.NET and Microsoft Azure. Even if your topic is not necessarily ML, this talk will teach you how to design and run heavy-load solutions in Azure.