AI-900: Microsoft Azure AI Fundamentals in a Weekend [2026]

Explore what artificial intelligence is, compare strong AI with narrow AI, and see how machine learning powers applications like spam filters, fraud detection, and image recognition.

Explore machine learning examples that predict image tags with confidence using Azure AI Vision, generate captions, and forecast house prices from historical data to train a model.

Compare machine learning with traditional programming, showing how rules-based design differs from learning from data to create models that make predictions. It also highlights data availability and updating models.

Explore three Azure AI approaches: pre-trained models, simple models with Custom Vision and AutoML, and complex models with Azure Machine Learning, while considering AI challenges, risks, and guiding principles.

Distinguish strong AI from narrow AI, see how machine learning learns from data, and preview Azure AI services, automated machine learning, custom vision, and Azure Machine Learning.

Explore different AI workloads such as content moderation, personalization, computer vision, natural language processing, knowledge mining, document intelligence, and generative AI.

Explore AI workloads across personalization, content moderation, computer vision with object detection, sentiment analysis, data mining, knowledge mining, and document intelligence like identity document extraction and generative AI content creation.

Explore pre-trained models and Azure AI services to quickly derive intelligence from images, text, speech, and documents, including Azure OpenAI, vision, face, language, and translation capabilities.

Learn image analysis, image classification, object detection, semantic segmentation, face detection, optical character recognition, and spatial analysis with Azure AI Vision, Azure AI Face, and Azure AI Document Intelligence.

Explore terminology through scenarios: image analysis, image classification, object detection with bounding boxes, semantic segmentation, OCR, face detection, and spatial analysis.

Explore how Azure AI Vision analyzes images using the Analyze Image and Read OCR APIs to generate captions, detect objects, tag images, and perform dense captions.

Explore azure ai services with microsoft foundry, an end-to-end platform providing access to OpenAI, Meta, Mistral models and tools to move from prototype to production with scalable compute.

Create an Azure account by submitting email, phone, personal details, and payment method; complete verification and temporary card authorization to sign up for a 30-day, 200 USD free credit trial.

Create a Microsoft Foundry resource in Azure to explore AI services like vision, language, and speech, using Foundry as a unified AI lifecycle with playgrounds, model fine-tuning, and deployment.

Explore vision plus document capabilities in Azure AI services, including object detection, image captioning, and dense captioning, using API keys to interact with models in the model catalog.

Explore the two-tier hub and project structure in Azure Foundry, enabling centralized control, shared data, credentials, and compute across projects, with separate hubs for finance and HR to isolate concerns.

Explore Azure AI Document Intelligence to extract text, structure, and key data with generic and prebuilt models for invoice, identity, check, pay stub, bank statement, contract, receipt.

Demonstrates Azure AI Document Intelligence in vision + document, using generic and prebuilt models to read text, identify layout, and extract invoice details such as amount due and billing address.

Explore Azure AI document intelligence scenarios by extracting text and layout from scanned documents using read and layout models, and apply prebuilt models for invoices, passports, checks, and more.

Explore Azure AI Face to detect and analyze faces, predicting age, emotion, and mask status. Utilize faceId, face lists, and person groups for detect, find similar, identify, and verify operations.

Explore Azure AI services through scenarios: detect faces and emotions with Azure AI Face service, generate image descriptions or tags with Azure AI Vision, and analyze PDFs with layout model.

Azure AI Services offers prebuilt, production-ready models with no training required, accessible via REST APIs. Use language-specific SDKs and flexible regional deployment to meet data residency needs.

Compare Foundry resources: multi-service resources grant a single key to access multiple AI services, while single-service resources provide access to one service, guiding your billing and app needs.

Explore how Azure Portal, Microsoft Foundry, and studios relate: Foundry manages AI resources on Azure, while studios offer service-specific web UIs like Vision Studio and Language Studio.

Access and authenticate AI services using the same Azure credentials, link resources in Foundry and Studio Portals, and use RESTful API endpoints with keys in request headers.

Explore Azure AI Language to analyze unstructured text with natural language processing, enabling sentiment analysis, key phrase extraction, text summarization, PII detection, and conversational language understanding.

Explore common text analysis techniques used in natural language processing, including text normalization, stop word removal, n-grams, stemming, and frequency analysis to reveal themes in text.

Explore the Azure AI language Text Analytics API to perform sentiment analysis, language detection, key phrase extraction, named entity recognition, and PII detection, with practical JSON responses and playground demos.

Explore Azure AI Language features such as named entity recognition, sentiment analysis, language detection, key phrases, and summarization, plus conversational language understanding, custom question answering, and orchestration workflows.

Master Azure AI speech service for real-time and batch speech-to-text, text-to-speech, and speech translation, including video subtitling, live transcripts, and voice assistance.

Explore Azure AI Translator service for real-time text translation across languages and document translation that preserves layout across formats like docx, presentations, html, and pdf, with synchronous and asynchronous options.

Explore natural language processing scenarios with Azure language services, including text analytics for sentiment, key phrases, and named entities, translator for translation, and speech service for speech-to-text and text-to-speech.

Explore Azure AI Bot Service for multi-channel chatbots with low-code or code-first options, integrating conversational language understanding and custom question answering, plus Azure AI Search and Azure AI Content Safety.

Stay ahead of technology by adopting the in28minutes learning pledge: 28 minutes a day for 28 days. Keep learning daily to evolve with technology and with frameworks, tools, and processes.

Explore how generative AI differs from artificial intelligence and machine learning, and how it creates new content by learning from examples.

Explore the difference between strong AI and narrow AI, focusing on ChatGPT. Learn to access ChatGPT, test prompts, and apply safeguards like avoiding sensitive data and fact checking.

See how ChatGPT aids coding, learning, and design with code generation, explanations, and API ideas, including a leap year function and todos REST API.

Leverage ChatGPT to explore technology and learn cloud engineering by generating a top-10 technologies list, covering cloud platforms, containers and orchestration, infrastructure as code, serverless, DevOps, and security.

Explore how generative ai requires huge data volumes and infrastructure to train statistical models, like GPT-3 trained on 500 billion words, that generate text, images, and code.

Explore how generative AI uses self-supervised learning to predict the next word from vast text data without explicit labels and learn from errors.

Explore how self-supervised learning powers text generation by predicting the next word, and learn to tune generation with temperature, top_k, and top_p parameters.

Discover how generative AI relies on deep learning to learn patterns from large data and generate new content across images, music, and text.

Learn how the loss function guides deep learning by comparing predicted outputs to actual results and minimizing the loss through feedback to keep predictions close to the actual output.

Learn to integrate generative ai into applications with the azure openai api, using GPT-4, GPT-3.5-turbo, and DALL-E to generate text, code, and images.

Learn how ChatGPT, based on the GPT language model, uses tokenization, stop words, and embeddings to understand language and leverage the transformer architecture.

Explore embeddings as 768-dimensional vector representations in high-dimensional space that capture semantic relationships and contextual information, enabling text similarity, NLP tasks, recommendations, clustering, and outlier detection.

Explore how embeddings and context drive the shift from RNNs to transformer architecture, with attention, encoder stages, embedding generation, and predicting the next token.

Tokenize text and embed tokens, apply positional encoding, then use multi-head attention to capture grammar, prepositional and contextual relations for context-aware next-word prediction.

Compare large language models and small language models by training data, parameters, and deployment, noting that large models require cloud or GPUs while small models deploy locally and fine-tune quickly.

Improve generative AI quality by mastering prompt engineering: set clear goals, add context, provide sources, refine prompts with system messages and retrieval augmented generation, and apply security and governance controls.

Understand foundation models and large language models, trained once and adaptable to many tasks across text and multimodal outputs. APIs enable apps like ChatGPT and DALL‑E to use them.

Explore the Azure AI Foundry model catalog to access open-source models as a service, streamlining model selection to deployment without provisioning infrastructure.

Deploy GPT-4.1 in Azure AI Foundry, use the Chat Playground to interact with the model, and configure system messages and safety prompts for tailored responses for marketing slogans.

Deploy the dall-e-3 image model in azure ai foundry and explore image generation in the images playground using prompts, with configurable size, style, and quality.

Explore Azure AI Foundry by browsing the model catalog and playgrounds, deploying GPT-4.1 and Dali, and using the management center to manage hubs, projects, models, endpoints, and connections.

Learn how NIST's AI risk management framework guides identifying, assessing, and monitoring AI risks, and how Microsoft's four-stage approach—map harms, measure harms, mitigate harms, operate with transparency—aligns with it.

Microsoft's four-stage approach to responsible generative ai maps, measures, mitigates, and manages potential harms, using red team testing, documented results, and phased rollout to safeguard content and compliance.

Explore Azure AI services to access GPT and DALL-E via the Azure OpenAI API and use the Azure AI Foundry model catalog, including retrieval-augmented generation, embeddings, attention, and transformer basics.

Embrace AI without fear and learn to consume, integrate, and use AI solutions proactively to build faster applications.

Set ambitious long-term goals to guide your learning across AWS, Azure, and Google Cloud, and translate them into concrete short-term milestones like deploying a full-stack app to each platform.

Build custom image models with Azure Custom Vision by creating and tagging image datasets, choosing between classification and object detection (multilabel or multiclass), and training a model to make predictions.

Train Custom Vision models with your images, perform classification or object detection, test predictions, and leverage hundreds of images with multiple tags for accuracy.

Understand features as inputs that power predictions and labels as the model’s outputs, and see how this distinction applies to house price, used car price, spam detection, and loan decisions.

Determine the right machine learning technique by examining labels, outputs, and the prediction goal. Distinguish supervised learning for regression and classification, and unsupervised learning for clustering.

Explore how to distinguish features and labels across supervised and unsupervised learning in regression, classification, and clustering scenarios, using house price and used vehicle price examples.

Obtain and clean data with sufficient variety, perform feature engineering to create features and labels, then train and evaluate models using different algorithms before deploying the best model.

Master core machine learning terminology, from training and evaluation to inference, datasets (features and label), and data splits (training, validation, testing) for model selection and production readiness.

Explore core machine learning stages and terminology, including data preparation, feature engineering, training, and inference, with model evaluation across training, validation, and testing datasets.

Explore how Azure Machine Learning simplifies creating and managing ML models, data, compute, and pipelines with automated machine learning and the Azure Machine Learning designer, plus publishing and monitoring.

Launch Azure Machine Learning Studio to build, train, evaluate, and deploy models using notebooks, Automated ML, and Designer; create a compute cluster and a diabetes dataset for pipelines.

Set up an automated ml training run for the diabetes experiment to train and tune a model using a target column y on a registered dataset, selecting regression.

Explore automated ML in Azure: run multiple algorithms, evaluate with normalized root mean squared error, deploy the best model to an Azure Container Instance, and test endpoints with real-time data.

Create an Azure machine learning pipeline in Designer by dragging datasets, data transforms, and regression algorithms onto a pipeline canvas, split data, train and score the model, and evaluate results.

Master classification model evaluation using confusion matrices and key metrics—true positives, true negatives, false positives, and false negatives—alongside accuracy, precision, recall, and F1 score.

Explore key model evaluation scenarios for classification and regression, using a confusion matrix and metrics such as accuracy, precision, recall, f1 score, mae, mse, and rmse.

Review essential Azure machine learning terminology, including Studio, Workspace, Designer, pipelines, datasets, modules, and compute options like compute instances, clusters, inference clusters, attached compute, and Databricks.

Explore building custom ML models in Azure with Custom Vision and Azure Machine Learning Studio, creating pipelines for training, scoring, and evaluation, then deploy via inference clusters and REST endpoints.

Teaching others enhances understanding and communication by simplifying complex ideas, conveys technology concepts clearly, builds confidence, expands knowledge, and fosters empathy, proving you learn twice through preparation.

Explore responsible ai principles, emphasizing data quality and representativeness, address bias, and apply six principles—fairness, reliability and safety, privacy and security, inclusiveness, transparency, and accountability.

Explore responsible AI principles through scenarios, identifying violations of fairness, reliability and safety, privacy and security, inclusiveness, transparency, explainability, and accountability in AI systems.

Review Azure AI services, including vision, document intelligence, language, face, speech, translator, content safety, search, Azure Open AI, machine learning, AutoML, and Foundry.

Identify key keywords across image processing, text analysis, and speech processing, from image classification and object detection to OCR, facial analysis, sentiment analysis, and translation with Azure AI services.

Identify essential generative AI keywords, including deep learning, embeddings, attention, and transformers. Learn how retrieval-augmented generation and foundation models like GPT, DALL-E, Gemini, and Llama power modern AI.

Explore the rebranding of Azure Cognitive Services to Azure AI Services, including Azure AI Search, Azure AI Vision, Azure AI Face detection, and more, with no API or SDK changes.

Discover deprecated Azure services and their replacements, including LUIS moving to Conversational Language Understanding, QnA Maker replaced by Custom Question Answering, and Form Recognizer joining Azure AI Document Intelligence Service.

Explore the ai-900 certification homepage, skills outline, and exam format for Azure AI Fundamentals; learn single- and multiple-answer question types and the 40-question, 60-minute exam.

Celebrate completing the AI-900 course and preparing for the Azure AI fundamentals certification while encouraging reviews, shared success stories in Q&A, and continued learning in Azure and machine learning.