This is how I practice for the GCP Machine Learning Engineer (MLE) certification.
To Start, there are a multitude of sources out there that can help you get started with practicing. Before we dive into the practical steps, I’d like to first highlight why this certification is useful.
Why MLE?
Machine Learning Engineering is one of the most coveted jobs in the industry. As machine learning (ML) is moving from research into production, there is a high demand on product oriented engineers who can translate this research into useable services and products. To succeed in the role as MLE, one should be proficient with software engineering as a base skill, and developed an understanding of general ML concepts/theory to be able to understand what models to train for different purposes, and how to optimize and fine tune the models to achieve the best possible outcome. One should be able to understand the performance metrics necessary to decide whether a model is suitable to deploy in for particular use case. It definitely helps to understand the business requirements in order to better adapt the models to fit the the needs and create value for the organization.
For Google Cloud Provider (GCP), however, there lies way more than knowing how to train and deploy an ML model in vacuum. GCP provides the context and environment necessary for handling your large datasets, and gives you the tools you need to train and deploy models. With GCP, you are looking at the entire value chain from data collection, storage, filtering and preprocessing data, to training, testing, deployment, CI/CD on Google Cloud, security, and so on. Essentially, GCP should be able to handle your entire process and give you the platform that allows you to be productive in your role.
Why GCP?
There are alternatives to GCP, most notably Amazon Web Services (AWS) or Microsoft Azure. When you start out as an MLE, typically, your org has already decided which cloud service provider to use. As I am writing this (2023-01-12) the market is divided as follows:
| Cloud Service Provider Ranked | Market Share |
| Amazon | ~ 48 % |
| Azure | ~ 16 % |
| ~ 4 % |