Artificial Intelligence #70: Cloud architectures for AI and ML  - AWS, Azure, GCP

Artificial Intelligence #70: Cloud architectures for AI and ML - AWS, Azure, GCP


Welcome to Artificial Intelligence #70

Firstly, a question: I am planning this leading to Christmas. What do you think? Who would you nominate??

Should we create a list / award for recognising the efforts of teachers of artificial intelligence?

As we launch our two courses,? Artificial Intelligence: Cloud and Edge Implementations. - and? University of Oxford Digital Twins course? ??we have been revamping cloud architectures in our teaching.

Anjali Jain and I have a session on Cloud architectures where we try to explore the full potential of the Cloud.?

Some notes re the below:

  1. Our belief is, if you understand one platform, you can understand the others also by thinking of parallels in functionality.?
  2. The cloud is evolving rapidly and hence the below will always be a moving goal post
  3. We emphasize AI due to the nature of the course
  4. We focus on AWS, Azure and GCP again due to the nature of our course

There are three segments to understand the cloud

1) Cloud architecture types?

2)?Cloud as a? platform to build services?

3)? Using? prebuilt cloud services??

In addition, you need to understand cloud design patterns?

Finally, I believe, Cloud platforms will evolve significantly driven by large language models(LLMs). Specifically, my 2022 prediction was: we are lively to see LLMs lead to low code applications with much of the AI functionality being implemented in the Cloud

But that's a discussion for another blog!

Cloud architecture types

In each case, we link AWS, Azure and GCP services

Infrastructure as a Service (IaaS) where you have full control and flexibility flexible and manage everything through VMs

EC2 ?

Virtual Machine ?

Compute Engine ?

Container as a Service (CaaS) - Deploy containers on a cluster that you configure and manage

ECS EKS ?

Azure Kubernetes Service (AKS) ?

Kubernetes Engine ?

Container as a Service (Serverless CaaS) - Containers on clusters managed and scaled by the cloud vendor

App Runner Fargate ?

Container Apps ?

Cloud Run

Platform as a Service (PaaS) - Canned app runtimes by the cloud vendor?

Elastic Beanstalk ?

App Service ?

App Engine ?

Function as a Service (FaaS) - For event-driven, stateless, short-lived services??

AWS Lambda

Azure Function

Cloud Function

Cloud as a? platform to build services?

Here, we describe cloud features that you can use to build your own services

again In each case, we link AWS, Azure and GCP services

Developing and deploying machine learning models?

SageMaker? ?

Azure Machine Learning ?

Vertex AI ?

Automated machine learning?

Amazon SageMaker? Autopilot

Azure ML - Automated Machine Learning

Vertex AI AutoML models

Preconfigured VMs for deep learning applications

Amazon EC2 P3 Instances

Azure Data Science ?

Deep Learning VM Images

Enterprise grade AI / ML deployment?

Tensorflow on AWS

Azure Databricks

TensorFlow Enterprise

MLOps,?

SageMaker for MLOps

Machine Learning Operations (MLOps)

Kubeflow

Data labeling

SageMaker Data Labeling

Azure ML - Data Labeling

Vertex AI Data Labeling Service

Explainable ML tools?

SageMaker Clarify

Azure ML – Model Interpretability

What-If Tool

Using? prebuilt cloud services??

Developing intelligent bots?

Alexa Skills Kit ?

Bot Framework ?

Dialogflow ?

Extracting information from text?

Amazon Textract ?

Form Recognizer ?

Document AI ?

Label image data

Rekognition Image ?

Azure Cognitive Services ?

Vision AI ?

Audio to text

Amazon Transcribe ?

Azure Cognitive Services ?

Speech To Text ?

text to speech

Amazon Polly? ? ?

Azure Cognitive Services ?

Text To Speech ?

Language translation?

Amazon Translate ?

Azure Cognitive Services ?

Translation AI ?

Label and categorize video content?

Amazon Rekognition Video

Azure Video Indexer ?

Video Intelligence API ?

Product recommendations

Amazon Personalize ?

Azure Cognitive Services? Personaliser ?

Recommendations AI ?

Advanced natural language processing:? sentiment analysis, key phrase extraction, language detection, and named entity recognition.

Amazon Comprehend ?

Azure Cognitive Services ?

Natural Language AI ?

understand user commands contextually.

Lex ?

Language Understanding LUIS ?

Natural Language AI

Cloud design patterns

Finally, we have Cloud design patterns?

1) Architecture Styles

Big compute

Big data

Event-driven architecture

Microservices

N-tier application

Web-queue-worker

?2) Choose an analytics solution

Analytical datastores

Analyticsand reporting

Batch processing

Stream processing

?3) Choose an AI/ML service

Cognitive services

Natural language processing

Machine learning

Machine learning tools

Compare MLflow and Azure ML

?4) Design principles

Design for self-healing

Make all things redundant

Minimize coordination

Design to scale out

Partition around limits

Design for operations

Use managed services

Use an identity service

Use the best data store

Design for evolution

Build for the needs of business

?

source Microsoft

If you want to study with us at the #universityofoxford , please have a look at our courses? Artificial Intelligence: Cloud and Edge Implementations. - and? University of Oxford Digital Twins course??

References?

Satish Chandra Gupta

Arockia Liborious?

Dr. PG Madhavan

Digital Twin maker: Causality & Data Science --> TwinARC - the "INSIGHT Digital Twin"!

2 年

Super-useful for a non-Cloud IT guy like me - mucho gracias! Would you consider adding under prebuilt cloud services, prebuilt cloud services: AWS IoT Greengrass IoT Streetwise IoT Core IoT TwinMaker Kinesis

Matthias Klein-Schl??l

Software für die Energiewende!

2 年

Very helpful link list. Thanks

Isabel Arens

EU Climate Pact Ambassador ?? Supporting sustainability innovation 2share@fainin @nekosystems @ideascanner

2 年

Ajit Jaokar please consider to nominate people who teach & apply ethics. There are justified threats towards AI development. One of them is Elon Musk who warns about the danger for society. Imagine AI used in war. I am concerned about the development without the right guidelines.

Herman Tiumentsev

Learner, AI Enthusiast, Machine Learning Researcher, Deep Learning Practitioner

2 年

????????????

要查看或添加评论,请登录

社区洞察

其他会员也浏览了