Google Cloud has added a Beta version of a new Professional-level certification to their available paths. The Professional Machine Learning Engineer certification exam will assess candidates' knowledge of machine learning practices and implementation on the Google Cloud Platform.
The new beta exam joins the seven other Professional-level certifications offered by Google Cloud Platform (GCP). The exam covers a variety of machine learning (ML) topics, oriented towards designing and implementing solutions using the TensorFlow deep-learning framework and GCP services. There are no hard pre-requisites, but Google recommends candidates have three or more years of experience with GCP. The exam fee is $120, and the certification is valid for two years. As with all GCP certifications, candidates who pass the exam will receive several benefits, including a sequentially numbered certificate, a digital badge, and the option to be listed in the GCP Credential Holder Directory. Google also claims that "almost 1 in 5" GCP certificate holders received a raise post-certification.
The exam covers the following areas:
- ML Problem Framing: translating business needs into ML requirements, including identifying the type of ML solution (e.g., classification or regression)
- ML Solution Architecture: identifying the proper GCP services to use for the ML solution
- Data Preparation and Processing: feature engineering and designing data pipelines
- ML Model Development: choosing model frameworks; training and testing models
- ML Pipeline Automation & Orchestration: using CI/CD to train and deploy models
- ML Solution Monitoring, Optimization, and Maintenance: production troubleshooting and performance tuning for models
Since early 2017, GCP has had a Professional Data Engineer certification that includes a machine learning component. However, the content was focused on "operationalizing" ML models, whereas the new exam covers the full ML lifecycle. The new exam's guide also calls out two technologies specific to Google's deep-learning framework TensorFlow: TFRecords and TensorFlow Transform. The exam otherwise appears to be framework-agnostic, though still oriented around using GCP services.
As was done with several of its other certifications, GCP is initially offering this new exam as a Beta. According to the certification documentation, Beta exams are "opened for a very short window, and are available sporadically." Historically, the Beta period for previous exams has averaged only a few months. A Beta exam is longer than other exams and is available in English only, but the registration fee is discounted by 40%. As with other exams, the Beta exam must also be taken at a dedicated test center.
To help measure the value of certification, Google recently commissioned an "independent third-party research organization" to survey 1,789 individuals who recently acquired a GCP certification. The survey results showed that the certification helped the holders with job search, promotions, and pay raises. According to the survey, nearly 20% received a raise, and more than 25% of holders "took on more responsibilities or leadership roles."
The other leading cloud providers, Amazon Web Services (AWS) and Microsoft Azure, also have certification programs similar to the Google Cloud program, including certifications focused on machine learning and AI. AWS announced its machine learning specialty exam in late 2018 and Microsoft announced their AI and data science certifications in early 2019. Although the Google Developer Network released a TensorFlow certification earlier this year, this is GCP's first ML-specific Professional-level certification. In a discussion on Reddit, one user noted:
Google is essentially putting a stake in the ground and lending its own definition of what [a machine learning engineer] is capable of and this will almost certainly influence the industry as a whole.
Unlike GCP's other certifications, the new exam has no practice exam available. However, the new certification's web page offers links to online training via Coursera and Qwiklabs, as well as in-person training through GCP's authorized training partners.