Author: Anne Lifton
Considering Google Cloud Platform (GCP) for your organization’s Machine Learning and advanced analytics needs? Are you hungry? Well, you’re in luck — whether you need data science, food, or both!
Here’s how deploying cutting-edge machine learning can feel more like a trip to a four-star restaurant than a trip to the grocery store.
In this article, we highlight a few of GCP’s core capabilities for building, training, and deploying machine learning in the cloud. GCP is an evolving cloud platform that has become a powerful and easy to use tool in the development of true full stack data science applications.
Level 1: Ingredients (AI/ML cloud platform management components or “raw foods”)
GCP can accelerate the deployment of machine learning systems by offering a wide variety of essential services, from Cloud Functions to providing discrete, timed operations, to Virtual Machines, hosted Kubernetes, API services, and logging and monitoring. All of these ”ingredients” are available in GCP to build a fully monitored system, backed with storage and data systems, as well as reporting services. It also offers the standard features of Kubernetes such as autoscaling and continuous deployment.
This system can also be governed via Google Data Catalog and can be deployed automatically using Google Deployment manager. Altogether, GCP provides for a true full-stack, governed data and prediction environment. By supplying the necessary ingredients GCP allows us to take those core ingredients and build extremely flexible solutions tailored to your needs.
[Read More: The Philosophy of Technology – A Q&A with Pero Dalkovski]
Level 2: Restaurant (pre-packaged services or “prepared meals”)
Continuing our analogy, the next level involves deploying machine learning via a package like BigQuery (BQ), which allows owners of extensive BQ databases to quickly and easily leverage the power of machine learning within their respective database. Solutions such as BigQuery use serverless architecture, allowing users to take advantage of machine learning capabilities at scale, without the need for infrastructure management.
Other ”prepared meals” examples include the amazing and powerful pre-trained services such as Google Translation Engine, Speech-to-Text, Vision, Recommender Services, Video Analysis (Video Intelligence), HR talent resources, and document analysis.
Each of these pre-trained services do require “going to the restaurant”, meaning the user does need to show up with data knowing how to use the specific services they are interested in. However, it offers some of the best in class “dishes” with access to some of the most cutting-edge prediction engines on the market.
Level 3: In-Home Chef (end-to-end Machine Learning Ops or “five course dinner”)
Finally, Google’s Vertex AI is the most deluxe of their fully automated, full-stack offerings. It has a suite of tools to assist in preparing a dataset for machine learning, it can automatically train machine learning models, and finally, it can create an API endpoint for making predictions on demand and in real time. This “in-home chef” equivalent can combine all of Google Cloud’s offerings into a single environment, allowing for easier management throughout the lifecycle of ML projects and, notably, across various model types and user expertise.
Whether you prefer to cook your meals from scratch, have your meals served on a silver platter, or anywhere in between, Google Cloud Platform has the tools you need to build, deploy, monitor, and experience the ROI your organization can achieve using Machine Learning. If you would like to accelerate your Machine Learning journey leveraging GCP or other cloud platforms, RevGen has the expertise to guide you — regardless of your starting point or culinary preference: raw ingredients, restaurant, or in-home chef.
To learn more about the other services we offer, check out our Technology Services site.
Anne Lifton is a Sr. Architect of Data Science and Machine Learning at RevGen. She has over 10 years of experience in building, deploying and managing the lifecycle of data science models across several industries and all three major cloud platforms.