Vertex AI

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 41

Intro to

Vertex AI
Start with Vertex AI
Product, Design and AI Expert
Adi Mazor Kario
● +20 in the tech industry
● Expert in Product, Strategy, Design, AI in
Voice/ Conversation and AI Products
● Book: Innovating Through Chaos
● Speaker & Podcaster (Invincible Innovation
Show)
● Course Creator - “Launch AI Products 101”
● Tech, Animals & Humans Lover
● Married, Mother of 4+ cat+ dog
This Talk
● Introduction to Vertex AI
● Managing Datasets
● Building and Training ML
● Explainable AI
● Deployment
● Gen AI
● Conversational AI
● Summary
Vertex AI is a managed ML platform to accelerate
experiments and deploy AI models
The platform combines data engineering, data science, and
machine learning engineering workflows, enabling teams to
collaborate using a common toolset and scale their
applications using the benefits of Google Cloud
Vertex AI
ML Workflow
Vertex AI Flow
Vertex AI Capabilities
Vertex AI Flow
Vertex AI Capabilities
From Data > Datasets> Model
- Vertex AI supports managed datasets for four data types: image, tabular, text, and videos.
- You can create and manage datasets in the Vertex AI dashboard page by clicking on
Datasets and then Create Project
Model Training
Vertex AI offers several options for model training
and deployment, including:
- AutoML, which allows users to train tabular, image,
text, or video data without writing code
- Preparing data splits
- Custom training
- provides complete control over the training
process
Model Training (AutoML - Image)
Model Training (AutoML- Tabular)
Model Training (AutoML- Text)
Text
Classification
Example
Model Training (AutoML- Video)
Data Labeling
Model Training (Custom)
Training
Example
Model Garden
The platform also includes a Model Garden for
- Discovering
- Testing
- Customizing
- Deploying Vertex AI and open-source models and assets
- Generative AI for access to Google's large generative AI models
Evaluate: Explainable AI
Explainable AI (XAI) is a set of processes and methods that allow human users to understand and
trust the results and output created by machine learning algorithms.
To build trust and confidence when putting AI models into production and helps adopt a
responsible approach to AI development
XAI is essential for ensuring that AI systems are:
- working as expected
- meeting regulatory standards
- allowing affected parties to challenge or change outcomes
- promoting end-user trust
- model auditability
- productive use of AI
https://cloud.google.com/vertex-ai/docs
Evaluate: Explainable AI

https://www.slideshare.net/martonkodok/vertex-ai-unified-ml-platform-for-the-entire-ai-workflow-on-google-cloud-250644015
Evaluate: Explainable AI
Some key aspects of explainable AI include:
- Accuracy: Ensuring that the system is working as expected and providing an accurate
explanation of the output1.
- Fairness: Ensuring that the AI system is fair and unbiased
- Transparency: Providing a clear understanding of the AI system's decision-making process1.
- Trust: Building trust between humans and AI systems by providing explanations for the
AI-generated outputs

https://cloud.google.com/vertex-ai/docs
Deploying
Seamlessly deploy models to endpoints, automate and scale projects, and send predictions
to the deployed models.

https://cloud.google.com/vertex-ai/docs
Deploying
To deploy a model to an endpoint in Vertex AI, you can use the Google Cloud console.
- In the Vertex AI section, go to the Models page, click the name and version ID of the model
you want to deploy, and select the Deploy & Test.
- Deploying a model to an endpoint associates physical resources with the model, allowing it
to serve online predictions with low latency.
- Another method to deploy a model is using the gcloud command line tool.
- The Vertex AI SDK for Python can be used to run the entire machine learning workflow in
Vertex AI Workbench, a Jupyter notebook-based development environment.

https://cloud.google.com/vertex-ai/docs
MLOps
Continuous delivery and automation pipelines for machine
learning systems

https://cloud.google.com/vertex-ai/docs
MLOps
After deploying models, Vertex AI provides end-to-end MLOps tools to automate and scale
projects throughout the machine learning lifecycle
These MLOps tools run on fully-managed infrastructure that can be customized and scaled based
on performance and budget needs.
- Users can also use the Vertex AI SDK for Python to run the entire machine learning workflow
in Vertex AI Workbench, a Jupyter notebook-based development environment
Other available interfaces include the Google Cloud Console, the gcloud command line tool,
client libraries

https://cloud.google.com/vertex-ai/docs
MLOps

https://cloud.google.com/vertex-ai/docs
Generative AI

https://cloud.google.com/vertex-ai/docs
AI Conversation
Vertex AI Search and Conversation offers a way for developers to:
- Build and deploy enterprise-ready generative AI-powered search and chat
applications
- Easy orchestration layer that abstracts away complexity
- Enterprise-ready data privacy and controls

https://cloud.google.com/vertex-ai/docs
AI Conversation- Key Features
Vertex AI Search and Conversation offers a way for developers to:
● Build and deploy search and conversational applications quickly
● Use a single platform for varying development needs
● Create personalized multi-modal and multi-turn experiences
● Enable transactions and connect data across your applications
● Ground responses in enterprise data and maintain full control

https://cloud.google.com/vertex-ai-search-and-conversation?hl=en
AI Conversation- Basics
The Vertex AI Conversation feature creates a special Dialogflow CX agent, called
a data store agent.
Vertex AI Search and Conversation parses your content and creates a virtual data
store agent powered by large language models. Your agent can then have
conversations about the content with your end-users.
You can specify your domain name and/or upload any number of documents, and Vertex AI Search and Conversation will index them. This
information is made available to your agent to access in real time. This unifies the power and accuracy of Google Search algorithms running
against your public or private data with Dialogflow's natural conversation capabilities.

https://cloud.google.com/vertex-ai-search-and-conversation?hl=en
AI Conversation- Use Cases
You can create experiences such as:
● Website agent: an agent that answers questions based on your website
contents
● FAQ agent: an agent that answers frequently asked questions
● Employee-facing agent: for answering help and support questions using
internal helpdesk or HR support documents
● More…

https://cloud.google.com/vertex-ai-search-and-conversation?hl=en
Conversation Data Store

https://cloud.google.com/generative-ai-app-builder/docs/agent-data-store
Conversation AI Steps
● Project Setup
● Create Data Store Agent
● Test Your Agent
● Deploy Your Agent
● Track Performance

https://cloud.google.com/generative-ai-app-builder/docs/agent-data-store
AI
Conversation
Example

https://cloud.google.com/vertex-ai/docs/samples/aiplatform-sdk-sentiment
-analysis?hl=en#aiplatform_sdk_sentiment_analysis-csharp
Vertex AI and More
Summary
Vertex AI is a comprehensive platform for the entire machine learning lifecycle, from
data management to model deployment and scaling. It offers various options for
model training and deployment, as well as tools for automating and scaling projects
throughout the machine learning lifecycle.

You might also like