Yacine AWS AI Practicioner - Section 4: AWS Managed AI Services

1. What are AWS AI Managed Services?

Pre-trained ML services. 

Responsiveness, redundancy, availability, regional coverage, performance.

Provisioned throughput: for predictable workloads, cost savings and predictable performance.

Token: pay for what you use.

2. What is Amazon Rekognition?

Amazon Rekognition is a computer vision service that analyzes images and videos.

It can detect objects, scenes, faces, text, and activities, and is commonly used for image moderation, identity verification, and video analysis. Create a database of “familiar faces”.

Use cases:

- Labeling

- Content Moderation

- Text Detection

- Face Detection and Analysis (gender, age range, emotions…)

- Face Search and Verification

- Celebrity Recognition

- Pathing (ex: for sports game analysis)

3. What is Amazon Rekognition Custom Labels?

You can label your training images and upload them to Amazon Rekognition. Amazon Rekognition creates a custom model on your images set. New subsequent images will be categorized the custom way you have defined.

ex. find your logo in social media posts, identify your products on stores shelves, …

4. What's the different between these 2 Rekognition features: Custom Labels vs Custom Moderation?

Custom Modertion:

A feature that lets you fine-tune the built-in Rekognition moderation model using labeled images to improve its accuracy for your platform.

To use when: When Rekognition moderation is mostly correct but needs tuning for your platform

- You train an adapter

- API usage stays the same

- You can only train existing moderation categories

- You cannot invent new labels

- It reduces false positives and false negatives

Custom Labels:

- You train a brand-new classifier

- You define any labels you want

- You call a different API

To use when: You need to detect things not covered by moderation categories

4.1 What is Amazon Rekognition Content Moderation?

Automatically detect inappropriate, unwanted, or offensive content. Integrated with Amazon Augmented AI (Amazon A2I) for human review.

Custom Moderation Adaptors

- Extends Rekognition capabilities by providing your own labeled set of images

- Enhances the accuracy of Content Moderation or create a specific use case of moderation

5. What are common use cases for Amazon Rekognition?

Face detection and comparison, content moderation, object and label detection, celebrity recognition, and unsafe content detection in images and videos.

6. What is Amazon Textract?

Amazon Textract extracts text and structured data from scanned documents. Unlike OCR-only tools, it understands forms, tables, and relationships between fields.

7. How is Textract different from basic OCR?

Textract extracts structured data such as key-value pairs and tables, while basic OCR only converts images into raw text without understanding layout or structure.

8. What is Amazon Comprehend?

Amazon Comprehend is a natural language processing (NLP) service that analyzes text to extract meaning, insights, and relationships. It identifies entities, key phrases, sentiment, language, and topics.

Named Entity Recognition (NER) - Extracts predefined, general-purpose entities like people, places, organizations, dates, and other standard categories, from text. Real-time or Async analysis.

9. What are common use cases for Comprehend?

Sentiment analysis, entity extraction, document classification, customer feedback analysis, and compliance monitoring.

- Language of the text

- Extracts key phrases, places, people, brands, or events

- Understands how positive or negative the text is

- Analyzes text using tokenization and parts of speech

- Automatically organizes a collection of text files by topic

- Analyze customer interactions (emails) to find what leads to a positive or negative experience

- Create and groups articles by topics that Comprehend will uncover

10. What is Amazon Comprehend Custom Classification?

You can use training data (in a S3 bucket) to train Comprehend Custom Classifier. Then Comprehend Custom Classifier organizes (tag) documents into categories (classes) that you define.

Supports different document types (text, PDF, Word, images...).

Real-time Analysis – single document, synchronous

Async Analysis – multiple documents (batch), Asynchronous

11. What is Amazon Comprehend Medical?

Amazon Comprehend Medical detects and returns useful information in unstructured clinical text (Physician’s notes, Discharge summaries, Test results, Case notes).

Uses NLP to detect Protected Health Information (PHI).

Analyze real-time data with Kinesis Data Firehose.

Use Amazon Transcribe to transcribe patient narratives into text that can be analyzed by Amazon Comprehend Medical.

12. What is Amazon Transcribe?

Amazon Transcribe converts speech into text using automatic speech recognition (ASR).

Automatically remove Personally Identifiable Information (PII) using Redaction.

Supports Automatic Language Identification for multi-lingual audio. 

Toxicity detection capabilities leveraging speech cues (tone, pitch) and text-based cues to determine toxicity.

13. What is Amazon Transcribe Improving Accuracy?

Allows Transcribe to capture domain-specific or nonstandard terms (e.g., technical words, acronyms, jargon…).

Custom Vocabularies + Customer Language Models (Learn the context associated with a given word).

14. What is Amazon Transcribe Medical?

A specialized version of Transcribe designed for healthcare. It understands medical terminology and generates clinical-quality transcripts.

15. What is Amazon Translate?

Amazon Translate is a neural machine translation service that converts text from one language to another in real time or batch mode.

16. What is Amazon Polly?

Amazon Polly converts text into lifelike speech using deep learning. It supports multiple languages, voices, and speaking styles.

Lexicons: Define how to read certain specific pieces of text (ex. AWS => “Amazon Web Services”)

Speech Synthesis Markup Language (SSML): Markup for your text to indicate how to pronounce it (ex. “Hello, how are you?”)

Voice engine: generative (+ expressive), long-form, neural (+ natural), standard (robotic)…

Speech mark: Encode where a sentence/word starts or ends in the audio. Helpful for lip-syncing or highlighting words as they’re spoken.

17. What is Amazon Lex?

Amazon Lex is a conversational AI service for building chatbots and voice assistants. It uses the same technology as Alexa to handle speech recognition and intent understanding.

Integration with AWS Lambda, Connect, Comprehend, Kendra

The bot automatically understands the user intent to invoke the correct Lambda function to “fulfill the intent”. The bot will ask the customer for ”Slots” (input parameters) if necessary.

18. What is Amazon Forecast?

Amazon Forecast is a time-series forecasting service that predicts future values using historical data. It is commonly used for demand forecasting and inventory planning.

19. What type of data does Amazon Forecast require?

Historical time-series data, optional related datasets (such as promotions or weather), and metadata describing the data structure.

20. What is Amazon Personalize?

Amazon Personalize is a real-time recommendation service.

It uses ML to deliver personalized recommendations based on user behavior and interactions.

Product recommendations, content ranking, personalized marketing, and user-specific suggestions.

21. What are Amazon Personalize Recipes?

Algorithms that are prepared for specific use cases. Recipes and Personalize are for recommendations.

ex. 

- Recommending items for users (USER_PERSONALIZATION recipes)

- Ranking items for a user (PERSONALIZED_RANKING recipes)

- Recommending trending or popular items (POPULAR_ITEMS recipes)

- Recommending similar items (RELATED_ITEMS recipes)

- Recommending the next best action (PERSONALIZED_ACTIONS recipes)

- Getting user segments (USER_SEGMENTATION recipes)

22. What is Amazon Kendra?

Amazon Kendra is an intelligent enterprise search service powered by ML.

It provides natural language search over structured and unstructured business content.

23. How is Kendra different from traditional search?

Kendra understands intent and context rather than relying only on keywords. It returns direct answers instead of just matching documents.

Kendra from user interactions/feedback to promote preferred results (Incremental Learning). Ability to manually fine-tune search results (importance of data, freshness, custom, …).

24. What is Amazon Mechanical Turk?

Crowdsourcing marketplace to perform simple human tasks (ex. labeling images). Integrates with Amazon A2I, SageMaker, Ground Truth…

25. What is Amazon Augment AI (A2I)?

Human oversight of Machine Learning predictions in production (employees, AWS contractors, Mechanical Turk). The ML model can be built on AWS or elsewhere.

26. What is Amazon Q?

Amazon Q is AWS’s generative AI assistant for builders and operators. It helps you build, code, debug, operate, and troubleshoot workloads on AWS.

27. What is Amazon Q Business?

Amazon Q is a generative AI assistant for businesses. It helps users query data, generate insights, and automate tasks using company-specific information.

ex. write a job offer to x position or write a social media post.

You can add Data Connectors to use data from 40+ data sources.

You can add plug-ins to connect to external services (Jira, SNOW, Zendesk, etc.).

You can use IAM Identity Center to authenticate users. It allows to use external Identity Provider. Users will only be able to access the files they have access to.

There is also Admin Controls (somewhat like Guardrails in Bedrock) to block words, topics, allow only certain data (external/internal), etc.

28. What is Amazon Q Apps?

Amazon Q Apps allow you to create Gen AI powered apps without coding by using NL and it’s based on your business data + you can add plug-ins (ex. Jira).

29. What is Amazon Q Developer?

Amazon Q Developer is a smart copilot for devs.

- Make changes to your account via Command Line Interface (CLI) in - Cloud Shell

- Write code and live code suggestion

- Security scan

- Implement features 

- Document code

- Write base code for new projects

- Integrates to IDE

ex. “List all my Lamba functions”

ex. “Do…”, Q writes a command for us

30. What is the difference between Amazon Q Business and Amazon Q Developer?

Amazon Q Business helps employees interact with company data and systems.

Amazon Q Developer assists developers with coding, debugging, AWS usage, and architecture guidance.

31.1 What is the difference between AWS Kendra and AWS Q Business?

Kendra is like Google with your enterprise data.

Q business is like ChatGPT with your enterprise data.

31. How does Amazon Q access enterprise data?

Amazon Q connects securely to internal data sources such as S3, databases, and SaaS applications. It respects IAM permissions and does not expose unauthorized data.

32. What is the difference between Rekognition and Textract?

Rekognition focuses on visual content analysis (objects, faces, scenes), while Textract focuses on document understanding (text, forms, tables).

33. What is the difference between Comprehend and Translate?

Comprehend extracts meaning from text (sentiment, entities), while Translate converts text from one language to another.

34. What is the difference between Lex and Bedrock?

Lex is designed for building conversational bots with predefined intents and flows.

Bedrock is a general-purpose generative AI platform using foundation models.

35. Are AWS managed AI services serverless?

Yes. These services are fully managed and serverless, meaning there is no infrastructure to provision or manage, and they scale automatically.

36. What does Amazon QuickSight allow you to do?

Amazon QuickSight is used to visualize your data and create dashboards with them.

36.1 What is Amazon Q for QuickSight?

Gen AI bot inside QuickSight that can answer questions and build chart using your data.

37. What is Amazon Q Developer in chat applications (ex Amazon ChatBot)?

You can deploy an a chat bot in a Slack or Microsoft Teams channel that knows about your AWS account.

Troubleshoot issues, receive notifications for alarms, security findings, billing alerts, create support requests.

You can access Amazon Q directly in the chat bot to accelerate understanding of the AWS services, troubleshoot issues, and identify remediation paths

38. What is Amazon Glue?

AWS Glue is an “ETL” (Extract Transform and Load) service used to move data across places.

Data is often spread across different places (DBs, data lakes, apps, logs).

Before analyzing it, the data usually needs to be collected, cleaned, formatted, moved... AWS Glue automates these steps.

Q in Glue allows to

Chat: 

- Answer general questions about Glue

- Provide links to the documentation

Data integration code generation:

- Answer questions about AWS Glue ETL scripts

- Generate new code

Troubleshoot:

- Understand errors in AWS Glue jobs

- Provide step-by-step instructions, to root cause and resolve your issues.

39. What is Amazon PartyRock?

GenAI app-building playground (powered by Amazon Bedrock). Allows you to experiment creating GenAI apps with various FMs (no coding or AWS account required).

UI is similar to Amazon Q Apps (with less setup and no AWS account required)

Basically vibe coding.

40. What is Amazon’s Hardware for AI?

GPU-based EC2 instances (P3, P4, P5, …, G3, G6, …)

AWS Trainium

- ML chip built to perform Deep Learning on 100B+ parameter models

- Trn1 instance has for example 16 Trainium Accelerators

- 50% cost reduction when training a model

AWS Inferentia

- ML chip built to deliver inference at high performance and low cost

- Inf1, Inf2 instances are powered by AWS Inferentia

- Up to 4x throughput and 70% cost reduction

Trn & Inf have the lowest environmental footprint

41. What is Amazon Connect?

Amazon Connect is a cloud platform for building and running customer support contact centers (calls, chat, messaging, and agent routing).

42. What is Amazon Q for Connect?

Amazon Q for Amazon Connect is an AI assistant that helps contact center agents (humans) answer customer questions and resolve issues faster during live conversations.

43. What is Amazon Transform?

AWS Transform is an AI tool that analyzes and automatically updates a company’s legacy software and infrastructure to modern cloud architectures. Uses Agentic AI.

44. What is Amazon CodeGuru?

Amazon CodeGuru is a tool that automatically reviews your code and suggests ways to fix bugs, improve performance, and reduce costs. Uses ML.

Main features:

1. Profiler. Identify which part of your code cost the most based on CPU usage.

2. Reviewer. Find bugs and critical issues.

45. What is the difference between CodeGuru and Q for Devs?

CodeGuru: analyzes your code and running applications automatically to detect bugs, inefficiencies, and performance issues. It gives specific, actionable suggestions.

Q for Devs: conversational AI assistant for developers. You ask questions, get explanations, or even get code snippets generated on demand.