AI-900 Microsoft Azure AI Fundamentals | Exam + Labs

Define cloud as internet-based provision of computing infrastructure with pay-per-use, then explain the shared responsibility model, contrasting customer and provider duties in on-premises versus cloud environments.

Pay only for what you use with a consumption based cloud model, turning capex into opex and enabling capacity that scales with demand; use Azure pricing calculator to anticipate costs.

Use the Azure pricing calculator to estimate costs for storage, ai vision image analysis, product recognition, and Ubuntu-based virtual machines in East US.

Learn about the three cloud service types: infrastructure as a service, platform as a service, and software as a service, and how responsibilities split between provider and you.

Explore Azure's physical hierarchy, from regions and data centers to availability zones and availability sets, and learn how resource deployment, pricing, and resilience vary across the world's cloud infrastructure.

Explore how every Azure resource sits in a resource group, within a subscription, and under management groups, then govern access, policies, and costs across the organization.

Navigate the Azure portal to manage resources and AI services, create resource groups, deploy computer vision, language, translator, OpenAI, and Azure Machine Learning resources, and monitor costs with Azure Advisor.

Explore the main AI workloads on Azure, including content moderation, computer vision, natural language processing, document intelligence, knowledge mining, and generative AI, and learn AI responsibility principles.

Examine how AI can go wrong—from biased responses to mass surveillance and manipulated public opinion—and learn how Microsoft's responsible AI principles in Azure guide safe, ethical use.

Explore Microsoft’s six responsible AI principles—fairness, reliability and safety, privacy and security, inclusiveness, transparency, and accountability—and apply them to AI systems.

Identify features and the label in a data set to train machine learning models. Learn four label types: two-class, multi-class, numerical, and no-label problems, and choose suitable algorithms.

Explore core machine learning concepts, contrast supervised learning with labels and unsupervised learning without labels, and apply regression, classification, multiclass, two-class, and clustering algorithms.

Explore the full machine learning lifecycle—from data sets and labels to supervised and unsupervised algorithms, training, testing, and deploying models via Azure Machine Learning in production.

Explore how Azure machine learning supports the end-to-end process—from data preparation and algorithm selection to training, evaluation, publishing via API, and ongoing maintenance—with automated ml and workspace setup.

Create an Azure Machine Learning workspace by provisioning a new resource group, storage account, key vault, and application, then access the Studio to begin modeling with the Titanic dataset.

Set up the azure machine learning compute layer by creating a virtual machine instance with idle shutdown and attach compute for training models in the ml studio, using Titanic dataset.

Take a quick tour of the Azure Machine Learning Studio portal to access notebooks, AutoML, designer, prompt flow, and assets for end-to-end model building and publishing.

Build a Titanic survival classifier in Azure Machine Learning Designer by importing a csv dataset, selecting features, splitting data, training a two-class boosted decision tree, and publishing an api.

Explore a regression workflow in Azure Machine Learning using a prebuilt dataset of vehicle features and prices to train, evaluate, and deploy a regression model that predicts vehicle prices.

Explore brain-inspired neural networks that learn by adjusting parameters across multiple layers to solve classification, regression, and clustering, with applications from self-driving cars to language translation.

Explore what computer vision is and how Azure offers pre-trained models to solve image-based problems. Learn object detection, image classification, face recognition, face analysis, and OCR.

Explore object detection, a computer vision task that identifies objects in images with bounding boxes and confidence scores. Practice using the Azure Computer Vision API in a quick lab.

Create and configure the Azure computer vision service in the portal, choose the standard pricing, and connect the endpoint and key in Visual Studio for object detection and OCR.

Experiment with object detection using the AI 900 computer vision API in Vision Studio, test images, and analyze JSON responses with bounding boxes, detected objects, and confidence scores.

Learn how optical character recognition enables computer vision to read and transcribe text from images, with practical uses from credit cards to receipts, menus, and translations.

Explore how the Azure OCR service, a vision service, reads text from images, transcribes it, and returns text with polygon, confidence, and JSON via the API.

Explore facial analysis with the Azure Face API, covering face detection to locate faces and landmarks, and face recognition by training a model to identify individuals.

Create a face API service in the Azure portal and link it to Vision Studio to enable face analysis. Detect faces and masks using the face service.

Explore image classification with Azure Custom Vision by training a model to distinguish Coca-Cola from Sprite, using a small dataset and pre-trained algorithms to classify images.

Train a custom vision model to classify images as healthy or near breaking pipes, then set up a custom vision service to label Coke or Sprite via the portal.

Train a custom classifier in the Azure Custom Vision portal by creating a project, uploading coke and sprite images, labeling them, training the model, and testing with samples.

Identify which Azure AI service handles each workload—ocr, object detection, face analysis, and custom vision—by creating the computer vision, face, or custom vision services and using their APIs.

Explore natural language processing, turning spoken and written language into usable data. Learn NLP workloads like transcription, sentiment analysis, keyphrase extraction, entity recognition, voice recognition, voice synthesis, and translation.

Learn keyphrase extraction in NLP to identify key phrases that describe text, enabling scalable insights from product reviews, social media, and free-text forms, using the Azure AI language service.

Master keyphrase extraction with the Azure language API via a rest call. Configure the endpoint, subscription key, region, and content type, and analyze English sample texts.

Identify entities in text with named entity recognition (NER) and map terms to categories like person, location, organization, skills, and emails, using Azure's finite entity set.

Perform an entity recognition lab that mirrors keyphrase extraction, using the api to identify entities across general and medical realms, including person, city, event, and organization.

Explore how sentiment analysis in NLP classifies text as positive, negative, neutral, or mixed using models. See Azure AI language service examples and lab exercises on customer reviews and tweets.

Change the URL to the sentiment endpoint, submit English text to the API, and obtain per-sentence sentiment with neutral, positive, and negative scores in this ai-900 lab.

Explore voice input operations: transcription converts audio to text, while voice recognition identifies speakers. Learn to use the Azure AI speech service for transcription labs and downstream NLP tasks.

Create a speech service in Azure, obtain an endpoint and keys, and link it to Speech Studio for real-time speech-to-text, text-to-speech, and a voice assistant.

Test real time speech to text in the speech studio using the Azure endpoint and English language, transcribing a sample passage about artificial intelligence.

Discover voice and speech synthesis, turning text into audio with NLP models; learn text-to-speech basics, uses like automated calls, and training models with your own voice.

Explore speech synthesis using text-to-speech with the Andrew voice in English (US), testing default, empathetic, and relieved styles, and applying AI to content creation.

Explore translation as a core NLP workload, translating text between languages with AI-powered models. Learn and use the translator service API in a hands-on lab to compare workloads and services.

Execute a curl call from the terminal to the Azure translator API v3.0, using the global endpoint and subscription key to translate English text to Spanish.

Identify how Azure nlp workloads map to specific services, including language service for keyphrase extraction, sentiment analysis, and entity recognition, speech service for transcription and synthesis, and translator for translation.

Generative AI is a type of AI based on deep learning that creates original content, such as text, images, audio, and code, with milestones like GANs, GPT-3, DALL-E, and ChatGPT.

Understand what language models are, how they use learned patterns to predict next words from prompts, and the difference between small and large language models used in generative AI.

Explore how language models use transformer architecture, tokens, vectors, and decoders to finish phrases. See how generative adversarial networks train a generator and a discriminator to improve image generation.

Compare large language models and small language models, their parameter scales, and use cases—from text classification and sentiment analysis to device deployment and general chat.

Discover how azure openai integrates openai technologies with microsoft, enabling gpt models, embeddings, dall‑e, and whisper via ai foundry, with copilot features and labs to build custom gpt apps.

Learn to interact with Azure OpenAI via Foundry, API, SDK, and Azure CLI to deploy and test a custom ChatGPT for your organization.

Configure the Azure OpenAI service in the Azure portal, create a resource, obtain endpoint and keys, and explore via the AI Foundry portal, linking workflows to NLP and computer vision.

Deploy and chat with Azure OpenAI in the AI foundry portal, compare data zone deployments, and run prompts through hands-on labs across ML, computer vision, NLP, and generative AI.

Generate images with the Azure OpenAI service via the Azure AI Foundry portal, using prompts to portray Santiago in Van Gogh and Andy Warhol styles, and dental logos.

Learn to generate code with Azure OpenAI service via chat in lab, including curl calls for sentiment analysis on Azure language API, a C double-linked list, and Python API examples.

Identify potential damages of generative AI, measure impact, and mitigate risks with content filters and deployment plans. Test for harmful outputs, enable incident response, and collect user feedback.