AI-900: Microsoft Azure AI Fundamentals with practice tests

Explore how artificial intelligence imitates human tasks across key workloads. Learn about machine learning, computer vision, natural language processing, document intelligence, generative AI, and knowledge mining.

Explore the six guiding principles of responsible AI from Microsoft—fairness, reliability and safety, privacy and security, inclusiveness, transparency, and accountability—for developing ethical AI-based applications.

Explore responsible AI questions from the AI-900 exam, including inclusiveness, reliability and safety, accountability, privacy and security, and transparency through explainable decisions.

Explore the fundamentals of computer vision, including image analysis, tagging, captioning, OCR, and object detection, using the Azure AI Vision service and Face service on Azure.

Explore Azure ai vision for image analysis, ocr, and object detection with bounding boxes and confidence scores; the demo covers creating the service in the Azure portal and vision studio.

Learn to call the azure ai vision api programmatically using powerShell, retrieve keys and endpoints, and use vision/v3.2 analyze to detect objects, captions, descriptions, and categories.

Explore key computer vision concepts like object detection with bounding boxes, distinguishing it from image classification, and apply OCR to read numbers on runners’ shirts for identification.

Explore how the Azure face service extends vision capabilities with advanced face detection, including bounding boxes, age and gender, and discuss the limited access policy for face recognition.

learn about face recognition operations in Azure AI Fundamentals, including one-to-many identification, one-to-one verification, finding similar faces, and grouping by similarity.

Explore natural language processing with the Azure AI language service, using language detection, keyphrase extraction, sentiment analysis, and named entity recognition.

Explore the Azure language service in the Azure portal and Language Studio, using features like sentiment analysis, key phrase extraction, language detection, and pre-built question answering.

Learn to call the text analytics api from a code environment, performing language detection, sentiment analysis, and key phrase extraction using language service resources and api endpoints.

Discover how to build a custom question answering feature with Azure language service. Create knowledge bases from FAQs, test responses, and use follow-up prompts for multi-turn interactions.

Explore conversational language understanding in the language service, using utterance, intent, and entity to power home automation chatbots and AI-based applications.

Contrast question answering and language understanding in the language service: question answering provides an answer from a knowledge base, while language understanding infers intent and reference entities to capture context.

Explore Azure's speech service for speech recognition and speech synthesis, and learn to fine-tune output with SSML to adjust pitch, tone, volume, and voice attributes.

Discover Azure AI Document Intelligence for automated data extraction from receipts, invoices, and business cards, using built-in models or custom training, and learn through portal workflows.

Explore the document intelligence service (formerly form recognizer) and its pre-built models that extract an invoice number and identify a retailer from a receipt.

Explore generative AI, including natural language generation, image generation, and code generation, powered by large language models; learn tokenization, embeddings, attention, and Azure OpenAI with GPT-4 and DALL-E.

Request access to the Azure OpenAI service for your company via the form, learn eligibility and responsible AI constraints, then explore the OpenAI Studio with GPT and DALL-E demos.

Explore how to set up and deploy Azure OpenAI services in the portal. Create a GPT-based chat deployment and configure system messages to steer responses.

Explore responsible AI practices for Azure OpenAI models aligned with Microsoft standards. Identify, measure, mitigate, and operate the four stages to manage harms in deployment.

Azure AI Foundry, the unified, one-stop shop formerly Azure AI Studio, blends OpenAI models with Microsoft and Meta models in a hub and project workspace for building AI applications.

Explore Azure AI Foundry and its models from OpenAI, Microsoft, Meta, Snowflake, and Databricks to perform text generation, object detection, image classification, and text-to-image.

Deploy a GPT model in Azure AI Foundry, from creating a hub and project to deploying the model and testing it in the playground with a system message.

Machine learning uses past observations to predict outcomes, using features as inputs and a label as the output, illustrated by predicting ice cream sales from weather data.

Explore supervised and unsupervised learning, including regression and classification under supervision, and clustering; learn features and labels with practical examples like ice cream sales and binary decisions.

An Azure machine learning model that predicts product quality distinguishes features as inputs and the label as what to predict; mass and temperature are features, quality is the label.

Create an Azure Machine Learning resource and workspace in the portal, train and test models with a 70/30 data split, using a heart disease dataset via a no-code web UI.

Explains how to create a data asset in Azure Machine Learning Studio, upload tabular csv data from local files to the workspace blob store, and validate it before training.

Demonstrates AutoML in Azure ML for heart data, showing automatic algorithm selection and feature engineering, with target column and automatic train-test split. Explains compute options including serverless advantages.