AI/ML and Analytics

A quick guide to AWS AI/ML services (SageMaker, Rekognition, Lex, etc.) and analytics services (Athena, Kinesis, Glue, etc.) for the CLF-C02 exam.

You Do Not Need to Build AI From Scratch

In the past, adding AI to an app required machine learning engineers, months of data collection, model training, and server deployments. AWS has compressed all of that complexity into simple API calls.

Face recognition, chatbots, voice transcription, and translation can all be added to your app without a data science team. For the exam, you only need to know which service fits which situation.

---

 

Each AI Service Is Like a Specialized Employee

Imagine a company with specialists for each task. AWS AI services are exactly like that — each one handles one specific job.

| Service | Job | Real-world example | |---------|-----|--------------------| | Amazon Rekognition | Security guard who recognizes faces and objects | Face-based door access, detecting inappropriate images | | Amazon Lex | Receptionist who answers questions | Website chatbot, automated customer service | | Amazon Comprehend | Analyst who reads and interprets text | Analyzing product reviews for sentiment | | Amazon Transcribe | Stenographer who types what people say | Converting meeting recordings into text | | Amazon Polly | Reader who speaks written text aloud | Converting news articles to audio | | Amazon Translate | Real-time interpreter | Multilingual support for a global app | | Amazon Textract | Clerk who reads and extracts data from documents | Automatically reading receipts or ID cards | | Amazon Kendra | Research specialist who finds answers in documents | "Does our internal knowledge base answer this question?" |

---

 

Amazon SageMaker — When You Want to Build Custom AI

The services above use AI that AWS already built. When your company needs a custom model trained on your own unique data, that is where Amazon SageMaker comes in.

SageMaker manages the entire lifecycle of a machine learning model — preparing data, training the model, evaluating performance, and deploying it to production — all in one platform. If you have ML expertise, use SageMaker. If you just need quick AI features, use the specialized services above.

---

 

Analytics Services — Getting Insights From Stored Data

Along with AI, analytics services also appear frequently in the exam.

| Service | Simple Description | When to Use | |---------|-------------------|-------------| | Amazon Athena | Run SQL on files in S3 without any server | Quickly analyze logs or CSV files stored in S3 | | Amazon Kinesis | Process data as it streams in, in real time | App click data or IoT sensor readings analyzed live | | AWS Glue | Clean, transform, and move data between places (ETL) | Consolidating scattered data from multiple sources | | Amazon QuickSight | Turn data into charts and dashboards | Creating visual reports for executives | | Amazon EMR | Process massive datasets using Hadoop or Spark | Handling hundreds of terabytes of big data |

---

 

Services People Often Confuse

Transcribe (speech to text) and Polly (text to speech) are opposites. Transcribe turns a recording into words. Polly reads words aloud.

Rekognition is for images and video only. To extract text from a document or scanned page, use Textract. To recognize a human face in a photo, use Rekognition.

Athena needs no server. You upload files to S3 and run SQL queries directly. There is nothing to set up or manage — it is fully serverless.

---

 

Exam Quick Reference

"Recognize faces or objects in photos and video" — Amazon Rekognition "Build a conversational chatbot" — Amazon Lex "Analyze text for sentiment or key phrases" — Amazon Comprehend "Convert speech to text" — Amazon Transcribe "Convert text to speech" — Amazon Polly "Real-time language translation" — Amazon Translate "Extract text from scanned documents" — Amazon Textract "Intelligent search inside documents" — Amazon Kendra "Build, train, and deploy custom ML models" — Amazon SageMaker "SQL on S3 data, no server needed" — Amazon Athena "Process real-time streaming data" — Amazon Kinesis "Serverless ETL, data transformation" — AWS Glue "BI dashboards and data visualization" — Amazon QuickSight "Massive big data processing with Hadoop or Spark" — Amazon EMR

Back to blog list