Vertex AI - The Data Science Platform for Practitioners

What is Vertex AI?

Vertex AI is a unified environment to accelerate experiments and deploy custom machine learning models. It offers a consolidated managed platform to work with custom code and pre-package models for experimentation, training, serving and operations following MLOps best practices.

Vertex AI unifies existing offerings into a single experience via SDK, API and UI for AI Platform Training, AI Platform Prediction, AutoML Tables, AutoML Vision, AutoML Video Intelligence, AutoML Natural Language, Explainable AI and Data Labelling. Besides, it integrates seamlessly with other Google Cloud services like BigQuery, DataFlow, Storage via an API or SDK.

Why should I care about Vertex AI?

For those already familiarized with AI Platform (Unified), Vertex AI is a rebranding of AI Platform (Unified).

Additionally, Vertex AI adds new operational features including Vertex Experiments to track, analyze and discover ML experiments for automated selection of best model candidates; Vertex Vizier which provides a sophisticated hyperparameter search to maximize model’s predictive performance; Vertex Pipelines which streamlines building and running ML pipelines to simplify MLOps pipelines automation.

Google has been a leader in offering on-demand and serverless solutions for advanced ML such as AI Platform and AutoML (Image, Natural Language, Video, Tables), because they offer an end-to-end ML lifecycle. However, there was a separation between those last two solutions making it very hard to put together a mixed solution.

We have heard from our customers that they’re interested in a ML Platform where they can manage datasets, models, retrain models using an automated ML pipeline, deploy model versions in a scalable way and split traffic depending on specific requirements. Then, if you are also interested in one of these features, give it a try on Vertex AI.

What is the key differentiator with previous and competitors solutions? Why should I not miss the chance to try it out?

Vertex AI is a platform for all data science teams and every team member, from a Data Analyst to a ML Engineer. It supports all sort of tasks in a ML workflow including managing all kinds of datasets (including tabular, image, video, documents); training a model using AutoML, common model architectures or your own custom ML model; adopting accelerated hardware, such as GPUs and TPUs, for all sorts of processing needs; leveraging an scalable hyper parameter tuning to optimize the model performance; requesting semi-automated data labelling jobs that lets you request humans to label your data; and managing models, all kind of artifacts and endpoints to integrate with your solution. In addition, Vertex AI supports all sorts of Deep Learning and Machine Learning libraries and frameworks in the market and any programming language.

Therefore, independent of the size and actual capacity of your Data Science practice, Vertex AI is an excellent match for any team size and experience. For instance, you can leverage a robust deep learning model training, evaluate the model performance, optimize it for deployment on any edge device by applying calibration and quantization techniques with the ease and convenience of working with the Google Cloud Web UI without the need to implement every single step taking place under the hood.

Where does it fit in/how can we weave it into our services portfolio?

Vertex AI sits at your application backend, giving you all the tools your Data Science team needs to experiment with different models, train the interesting model candidates and deploy machine learning models. Your application will communicate with the ML models via the endpoints available via HTTP or GRPC protocols to get predictions and expose those insights to the end users. Either those end users may need offline/on-device deployments or could serve web-based and mobile applications.

If you already use the AI Platform, it only takes one to follow the Migration Plan described in the Vertex AI documentation for the needed features. You can also count on Searce expertise to migrate your workloads from AI Platform to Vertex AI.

However, if you want to start fresh and implement or shift some of the ML workloads to Vertex AI; first, check the documentation to see what features are currently available, next reach out to me or any member of the Searce team to help you on the journey.