Machine Learning

ML Services Overview

AWS provides a broad set of AI and ML services. For the Cloud Practitioner exam, you need to know what each service does (not how to use it). Think of these as pre-built ML capabilities you can plug into your applications.

Amazon Rekognition

• Find objects, people, text, scenes in images and videos using ML.
• Facial analysis & recognition — Compare faces against a database, detect emotions, age range, gender.
• Content moderation — Detect inappropriate, unwanted, or offensive content in images/videos.
• Celebrity recognition — Identify celebrities in images/videos.
• Text in images — Detect and extract text from images (like license plates, street signs).
• Pathing — Track the path of people in a video (e.g., for sports analysis).
• Use cases: user verification, people counting, content moderation for social media, celebrity recognition.

Amazon Transcribe

• Automatically convert speech to text using deep learning (Automatic Speech Recognition / ASR).
• Supports automatic language identification for multi-lingual audio.
• Automatic subtitle generation for videos.
• Can redact PII (Personally Identifiable Information) using Vocabulary Filtering.
• Use cases: transcribing customer calls, subtitles, meeting notes, medical transcription (Amazon Transcribe Medical).

Amazon Polly

• Turn text into lifelike speech using deep learning (Text-to-Speech / TTS).
• Supports multiple languages and voices.
• Uses SSML (Speech Synthesis Markup Language) for advanced customization: pauses, emphasis, pronunciation, whisper, Newscaster style.
• Lexicons — Customize pronunciation of specific words (e.g., "AWS" as "Amazon Web Services").
• Use cases: accessibility features, voice-enabled applications, announcements, audiobook generation.

Amazon Translate

• Natural and accurate language translation using neural MT.
• Supports 75+ languages.
• Allows localization of content for international users: websites, applications.
• Can translate large volumes of text efficiently.
• Use cases: multilingual user experiences, translate user-generated content, real-time communication translation.

Amazon Lex & Amazon Connect

Amazon Lex

• Build conversational bots / chatbots using the same technology that powers Alexa.
• Automatic Speech Recognition (ASR) — Converts speech to text.
• Natural Language Understanding (NLU) — Recognizes the intent of the text/caller.
• Helps build chatbots, call center bots, voice-based interfaces.

Amazon Connect

• Cloud-based virtual contact center — Receive calls, create contact flows.
• No upfront payments, 80% cheaper than traditional contact center solutions.
• Integrates with other CRM systems or AWS services.
• Can integrate with Amazon Lex for intelligent call routing.

Amazon Comprehend

• Fully managed Natural Language Processing (NLP) service.
• Uses ML to find insights and relationships in text:
• Sentiment Analysis — Positive, negative, neutral, mixed.
• Key Phrase Extraction — Automatically extract key phrases.
• Language Detection — Identify the dominant language.
• Entity Recognition — Identify people, places, events, organizations.
• Topic Modeling — Group documents by topics.
• Amazon Comprehend Medical — Extract medical information from unstructured clinical text (physician notes, clinical trial reports, patient health records).
• Use cases: analyze customer reviews, social media sentiment, support tickets, legal document analysis.

Amazon SageMaker

• Fully managed service for developers and data scientists to build, train, and deploy ML models.
• Covers the entire ML workflow: labeling, building, training, tuning, deploying, monitoring.
• Provides Jupyter Notebooks for development.
• Supports built-in algorithms and custom algorithms.
• SageMaker Studio — Integrated development environment (IDE) for ML.
• Auto Scaling for deployed models.
• This is the most complex ML service, meant for data scientists who want to build custom models.

Amazon Forecast

• Fully managed ML-based forecasting service for time-series data.
• Up to 50% more accurate than traditional methods.
• Use cases: product demand planning, financial planning, resource planning, weather forecasting.
• You provide historical time-series data, and Forecast automatically builds a model.
• No ML experience required.

Amazon Kendra

• ML-powered enterprise document search service.
• Allows natural language search — you can ask questions in plain English.
• Extracts answers from documents (PDF, HTML, Word, FAQs, PowerPoint, etc.).
• Learns from user interactions and feedback to promote preferred results (incremental learning).
• Can index data from various sources: S3, RDS, SharePoint, Salesforce, ServiceNow, OneDrive, and more.
• Use cases: internal knowledge base search, IT support, HR FAQ, legal document search.

Amazon Personalize

• Build real-time personalized recommendations using ML.
• Same technology used by Amazon.com for product recommendations.
• Integrates into websites, apps, SMS, email marketing.
• Use cases: personalized product recommendations, customized direct marketing, customized search results.
• You provide data via S3 or API (user interactions, item catalog, user profiles), and Personalize builds the model.

Amazon Textract

• Automatically extract text, handwriting, and data from scanned documents using ML.
• Can extract from: forms (key-value pairs), tables, and unstructured text.
• Works with PDFs, images (JPEG, PNG), and TIFF files.
• Goes beyond simple OCR — understands the structure and context of documents.
• Use cases: invoice processing, tax form processing, ID document extraction, medical records, financial statements.

Developer ML Tools

Amazon CodeGuru

• ML-powered service for automated code reviews and application performance recommendations.
• CodeGuru Reviewer — Automated code reviews: finds critical issues, security vulnerabilities, hard-to-find bugs.
• CodeGuru Profiler — Provides application performance recommendations: identifies expensive lines of code, CPU utilization analysis, heap summary.
• Supports Java and Python.

Amazon CodeWhisperer

• AI-powered code companion that generates real-time code suggestions as you type.
• Trained on billions of lines of code (open-source and Amazon code).
• Supports multiple languages: Python, Java, JavaScript, TypeScript, C#, and more.
• Provides security scanning to find vulnerabilities in generated code.
• Integrates with popular IDEs: VS Code, JetBrains, AWS Cloud9, Lambda console.
• Reference tracking — flags suggestions that resemble open-source training data.

Quick Reference: Which Service to Use?