Exam Topics

The successful performance in the Google Professional Machine Learning Engineer certification test requires a good comprehension of its topics. The exam syllabus consists of six sections that are described below:

• Framing Problems Related to Machine Learning

  Within this subject area, the candidates should be capable of translating business challenges into the Machine Learning use cases. They should also possess the skills in determining the Machine Learning problems, identifying the business success criteria, as well as defining risks to the feasibility of the Machine Learning solutions.

• Monitoring, Optimizing, and Maintaining Machine Learning Solutions

  This objective evaluates the competency of the applicants in monitoring and troubleshooting the Machine Learning solutions. The individuals should also be able to tune the performance of Machine Learning for training and serving in production. This involves the ability to optimize and simplify the input pipeline for training as well as knowledge of the simplification techniques.

• Developing Machine Learning Models

  To answer the questions related to this section, the learners should know how to build, test, and train models. They should also possess the skills in scaling model training as well as serving, including distributed training and scaling prediction service (for instance, containerized serving, AI Platform Prediction, etc.).

• Architecting Machine Learning Solutions

  Here the examinees need to demonstrate their proficiency in designing reliable, scalable, and highly available Machine Learning solutions. Besides that, the test takers need to be capable of selecting the proper Google Cloud hardware components, including evaluating accelerator and compute options (for example, CPU, TPU, GPU, edge devices). Lastly, they need to have the expertise in designing an architecture that meets the security concerns across the industries/sectors.

• Designing Data Preparation & Processing Systems

  The aim of this topic is to measure the individuals’ skills in exploring data (Exploratory Data Analysis). This involves their understanding of visualization, statistical fundamentals at scale, data quality & feasibility evaluation, as well as data constraint establishment. It also evaluates the ability of the test takers to build data pipelines, in particular, organize and optimize training datasets, validate data, handle missing data, handle outliers, etc. You should also know how to create the input features (feature engineering). This envisages the familiarity with encoding structured data types, feature selection, class imbalance, feature crosses, transformations, and more.

• Automating & Orchestrating Machine Learning Pipelines

  This module encompasses one’s competency in designing & implementing training pipelines. This includes your ability to define the components, triggers, parameters, and compute needs; understanding of the orchestration framework; familiarity with the multi-Cloud or hybrid strategies; knowledge of system design involving the TFX components/Kubeflow DSL. The candidates should also possess the skills in implementing serving pipelines, including serving (online, caching, batch), testing for target performance, configuring trigger & pipeline schedules, among other skills. Apart from that, this part requires the students’ expertise in tracking & auditing metadata.

For more info read reference:

Google Web Services Website

Reference: https://cloud.google.com/certification/guides/machine-learning-engineer