Introduction to Building Scalable Machine Learning Models

Machine learning is a forever-evolving space. In the early days, the focus was on implementing the historic algorithms using the computing power that was previously unavailable. Currently, ML has expanded across multiple avenues of interest: Being able to scale models with the ever-increasing quantities of data, the ability to package models up so that they can fit into an MLOps framework, how well tracked and logged the models are, how explainable they are, not to mention techniques to squeeze the best performance out of the models. All these aspects are rarely covered all at once. This pre-con aims to take all these key topics and deliver them over the Databricks platform, one of the leading tools in the industry for data manipulation, machine learning and distributed computing. It utilises a powerful Spark engine under the hood whilst presenting an easy-to-use bespoke notebook for the front end.

This session calls upon experience gained over the years by experienced machine learning practitioners to deliver a condensed, concise course to help fast-track any machine learning novice towards mastery.

The pre-con will cover building real-world models in Azure, further techniques for robust model building, the latest craze in ML known as feature stores, advanced model tracking and key considerations when scoring models. The pre-con will contain talks, coupled with follow-along labs giving an element of hand-holding with some more tricky tasks thrown in. The sessions will also explore real-world use cases that are rife in the business world.
Wednesday 15 March 2023