AWS Cloud Quest — Machine Learning Role
Hello Medium Readers, after two weeks without posting, I’m back bringing my review about AWS Cloud Quest for the Machine Learning Role. As I did in my last post about the Data Analytics Role, I’ll start with my review then the services that I used, finishing with the Assignments developed by me.
I was positive surprised with the Assignment’s level of this Role, of course, the main idea is to present and teach you about the Machine Learning services available in AWS, although that, almost every assignment you can access the SageMaker Studio, which I cover in a post last year, and have contact with the code of the application.
I had contact with code to train and deploy Reinforcement Learning frameworks, using transfer learning for TensorFlow Model, how to train and deploy an ML model using AWS SageMaker infrastructure. Additional to that, the availability and covering of LLM subject was nice, in one of the assignment I fine-tuned a model to generate an image based on the Diffusion Model, available in HuggingFace, you can see the image generated below.
This CloudQuest was even more specialized and deeper than the previous one that I shared with you in my last post about Data Analytics. Now it’s over the Role more closely related to ML or DA, so I’m still deciding if I’ll face off the Solution Architect CloudQuest or if I start another training in AWS Skill Builder to go deeper into ML in AWS.
At the end, I strongly continue to suggest you to take a special look into AWS Cloud Quest, I think it pays its price. Although this role wasn’t too much repetitive like the Data Analytics, what I really liked, it covered a broader range of applications and services.
Take a look into new Badge :D
Machine Learning Role — Assignments
In the following sections, I’ll outline the 14 new assignments I undertook, accompanied by images representing the proposed solution architecture for each assignment. Additionally, it’s provided step-by-step guidance on solving each problem, detailing the AWS services utilized to achieve the desired outcomes.
It’s worth noting that the episodes are presented in the order in which I completed them. While some assignments may have dependencies on others, the sequence can be customized based on individual preferences and requirements.
Episode — Image and Video Analysis
- AWS Service: Amazon Rekognition
- Tasks: Detect objects in images using Amazon Rekognition and configure a Lambda function to be invoked by an S3 event.
Episode — Text-to-speech
- AWS Service: Amazon Polly
- Tasks: Synthesize text-to-speech using Amazon Polly, explore the Polly API, and automate text-to-speech synthesis using Lambda and S3.
Episode — Extract Text from Docs
- AWS Service: Amazon Textract
- Tasks: Extract data from documents using the Textract console, configure a Lambda function with provided code, and set up an S3 event notification.
Episode — Speech-to-Text
- AWS Service: Amazon Transcribe
- Tasks: Convert audio files to text using Amazon Transcribe and configure a Lambda function with an S3 trigger event.
Episode — Set Up an ML Environment
- AWS Services: Amazon SageMaker, Lambda, S3
- Tasks: Deploy an AI/ML learning environment using SageMaker, explore SageMaker Studio IDE, and walkthrough a sample ML project.
Episode — Spy Drones Detection
- AWS Service: Amazon SageMaker
- Tasks: Utilize SageMaker Studio for machine learning, deploy, train, test, and evaluate a supervised ML algorithm, deploy a SageMaker inference endpoint, and use Lambda for real-time predictions.
Episode — Anomaly Detection
- AWS Services: Amazon SageMaker, Lambda
- Tasks: Understand unsupervised learning algorithms, train and deploy an ML model for anomaly detection using Random Cut Forest algorithm, and integrate Lambda with a SageMaker endpoint.
Episode — Bring Your Own Model (BYOM)
- AWS Services: Amazon SageMaker, Lambda, S3
- Tasks: Import code from S3 to SageMaker Studio, deploy a model using a Jupyter notebook, and invoke the inference endpoint using Lambda.
Episode — TensorFlow and Compute Vision
- AWS Service: Amazon SageMaker
- Tasks: Import and run sample lab code in SageMaker Studio, train and deploy ML models using TensorFlow, and deploy models in SageMaker endpoints.
Episode — Get Home Safe
- AWS Service: Amazon SageMaker
- Tasks: Utilize SageMaker Studio notebook to train models for safety in Gym environment and deploy trained models in SageMaker.
Episode — Introduction to Generative AI
- AWS Services: Amazon SageMaker, Lambda, API Gateway, CloudFront, S3
- Tasks: Deploy models on SageMaker endpoints, use Lambda to invoke endpoints, and review/test applications.
Episode — Fine-Tuning an LLM on Amazon SageMaker
- AWS Services: Amazon SageMaker, Lambda, API Gateway, CloudFront, S3
- Tasks: Fine-tune models using SageMaker notebook instance, deploy fine-tuned models to an inference, and test with sample app.
Episode — Text-to-Image Creation Using Generative AI
- AWS Services: Amazon SageMaker, Lambda, API Gateway, CloudFront, S3
- Tasks: Deploy LLM on SageMaker endpoint, fine-tune LLM with image datasets, and deploy fine-tuned model on SageMaker endpoint.
Episode — Chatbots with a Large Language Model (LLM)
- AWS Services: Lambda, Lex, SageMaker
- Tasks: Understand LLM, create chatbots using Lex, and deploy pre-trained LLM integrated with Lex using Lambda.