Latihan Azure AI Engineer AI-102

Azure AI Foundry brings together AI services, models, and tools in one platform. DNS, databases, and CDNs are different.

Azure AI Services provide prebuilt AI APIs. VMs, storage, and DNS are infrastructure services.

Responsible AI is a set of principles that guide safe AI. Firewalls, models, and storage are not principles.

Content Safety filters harmful or unsafe content. Storage, translation, and VMs are different services.

Language resources contain endpoint and key configuration for Azure AI Language. Tables, DNS, and firewalls are different.

AI hubs organize projects and shared resources. Models, datasets, and load balancers are components.

Projects group assets for a solution. VMs, buckets, and DNS zones are different.

The model catalog lists deployable models. Databases, firewalls, and CDNs are not model catalogs.

Endpoints provide URLs for calling AI services. Containers, networks, and databases are different.

AI Foundry builds apps, AI Services provide APIs, responsible AI guides safety, and content safety filters harm.

Vision analyzes images. Language handles text, Speech handles audio, and Search indexes content.

OCR extracts printed or handwritten text from images. Translation, speaker recognition, and audio classification are different.

Custom Vision builds custom image classifiers and object detectors. Speech, Language, and Search are different.

Face detection finds faces and their attributes. Translation, speech, and embeddings are different.

Facial recognition matches faces to identities. Object detection, OCR, and captions are different.

Image classification assigns labels to images. Speech, translation, and reports are different.

Object detection finds objects and draws boxes around them. Speaker recognition, key phrases, and filters are different.

Captions describe image content in text. Folders, versions, and DNS records are different.

Thumbnails are small versions of images. Tables, speech models, and routes are different.

Vision analyzes images, OCR extracts text, Custom Vision detects custom objects, and Face detects faces.

Language provides text analytics. Vision analyzes images, Speech handles audio, and Search indexes content.

Sentiment analysis evaluates text tone. Faces, audio, and thumbnails are different.

NER finds named entities in text. Speaker recognition, object detection, and captions are different.

Key phrase extraction highlights important text. OCR, speech translation, and embeddings are different.

Translator converts text between languages. Vision, Speech, and Search are different.

Document Intelligence extracts fields and tables from documents. Speaker, speech, and VM tasks are different.

Question answering returns answers from defined content. Image generation, faces, and reports are different.

CLU extracts intents and entities from conversation. TTS, image classification, and content filters are different.

PII detection finds personal data in text. Faces, thumbnails, and audio translation are different.

Language analyzes text, Translator translates, Document Intelligence extracts data, and question answering returns answers.

Speech provides speech-to-text and related audio capabilities. Vision, Language, and Search are different.

TTS converts text into audio. OCR, object detection, and embeddings are different.

Speech translation converts spoken input into another language. Captions, image classification, and filters are different.

Custom speech adapts models to domains and vocabulary. Databases, DNS, and load balancers are different.

Speaker recognition identifies people from voice. Face detection, OCR, and thumbnails are different.

Pronunciation assessment scores how well someone pronounces words. Translation, images, and filters are different.

SSML controls how text is spoken. Storage, DNS, and firewalls are different.

Speech endpoints provide URLs for speech operations. Models, datasets, and routes are different.

Real-time transcription converts live audio to text. Document translation, object detection, and reports are different.

Speech-to-text transcribes, TTS speaks, custom speech adapts to domains, and speaker recognition identifies people.

Azure OpenAI provides OpenAI models through Azure. DNS, databases, and CDNs are different.

Prompts are inputs that guide model behavior. Outputs, folders, and routes are different.

Completions are model-generated outputs. Prompts, tables, and DNS records are different.

Embeddings encode text as numeric vectors for similarity and search. Audio, images, and firewalls are different.

Tokens are text units used by models. VMs, storage, and DNS are different.

Temperature controls randomness. Capacity, speed, and model size are different.

Max tokens caps output length. Model count, storage, and network are different.

System messages define assistant behavior. User questions, outputs, and databases are different.

Content filtering blocks harmful outputs. Storage, DNS, and load balancers are different.

GPT models generate text, embeddings are vectors, tokens are text units, and temperature controls randomness.

Deployments host models behind endpoints. Tables, DNS records, and firewalls are different.

Managed endpoints are hosted and operated by Azure. Networks, storage, and VMs are different.

Containers package AI services for portable deployment. DNS, databases, and CDNs are different.

Azure Monitor collects logs and metrics. DNS, CDN, and Storage are different.

Application Insights monitors app performance. Databases, firewalls, and models are different.

Quotas limit resource usage. Models, datasets, and DNS records are different.

API keys authenticate service calls. Versions, containers, and routes are different.

Model versions identify releases. Storage, firewalls, and DNS zones are different.

Evaluation measures output quality. Deletion, DNS, and storage are different.

Deployments host models, containers provide portable runtimes, Monitor collects metrics, and API keys authenticate.