End-to-End Data & AI: Microsoft Fabric + Azure OpenAI

Microsoft Fabric unifies data engineering, data lakes, data science, analytics, and AI workloads in a single SaaS platform, enabling a one data lake, governance with Purview, and copilot for fabric.

Explore data analytical engines in Microsoft Fabric, including Apache Spark and SQL analytics endpoint, and learn how startup pools and custom pools scale workloads in the lake house.

Explore the Microsoft Fabric portal, create a workspace and lakehouse, and learn to query data with the SQL analytics endpoint using Delta Lake and Parquet storage, plus Power BI integration.

Transform data with Apache Spark and Transact-sql in a Microsoft Fabric lake house, loading CSVs, converting to parquet and delta, and running SQL via notebooks or the SQL analytics endpoint.

Learn to use the default semantic model in a lakehouse to build Power BI reports from a delta table, creating charts of item and sum of quantity.

Explore data lake house concepts, including OneLake and Delta Lake ACID features, showing how Fabric unifies data lake, data warehouse, and workspaces into a single source of truth.

Explore how OneLake Explorer visualizes lake house data inside Microsoft Fabric, showing workspaces, files, and delta parquet format tables in your local file explorer.

data warehouse provides high performance for structured, complex queries and simpler dimensional models, enabling migration of gold layer aggregates from the data lake house to power fast BI reports.

Import delta table from the lake house into a new data warehouse, create a sales schema, build a fact table, and load 2019 data via a stored procedure.

Explore semantic models in Microsoft Fabric and Power BI, learn how data lake house and delta tables form a data warehouse, and map facts to dimensions to build dashboards.

Build a facts and dimensions semantic model from a sample data warehouse. Establish relationships and generate Power BI dashboards and reports.

Explore how medallion architecture organizes raw data into bronze, silver, and gold layers in a data lake house, delivering acid-compliant delta tables for BI and AI workflows.

Implement medallion architecture in a fabric data lake house, moving a Covid CSV from bronze to silver to gold, then build a Power BI dashboard.

Explore how data flows transform data in a no-code to low-code environment, then orchestrate them with fabric pipelines to automate ETL from sources to data warehouses and Power BI.

Create a data flow from a raw CSV URL, publish a table to a data lake house, then drive a Power BI dashboard via SQL analytics endpoint.

Learn to build custom AI and ML workloads in Microsoft Fabric, with data in a lakehouse inside a Fabric workspace, tracked by MLflow and deployed via REST endpoints.

Explore basic machine learning concepts on Azure Databricks, including features and target columns, train-test splits, mean squared error, and the bias-variance trade-off to avoid overfitting and underfitting.

Explore linear algorithms from linear regression to ridge and lasso shrinkage, with hyperparameter tuning using MLflow. Learn how logistic regression handles classification with the sigmoid function for binary outcomes.

Explore non-linear machine learning with decision trees, starting from root to leaf nodes, and optimize information gain and tree size to prevent overfitting while mastering bagging, boosting, and random forest.

Build a diabetes prediction model using a csv training dataset and logistic regression in a fabric workspace with a binary 0/1 outcome. Apply vectorized features and MLflow logging.

Deploy a diabetes predictor inside a fabric workspace, fetch diabetes testing csv via a data flow, and run a model inference pipeline to produce a diabetes final table with predictions.

Use MLflow in Microsoft Fabric to log model metrics, hyperparameters, and results from a diabetes logistic regression experiment, compare runs, and register the best model.

Explore hyperparameter tuning in a Microsoft Fabric workspace using the Hyperopt library to optimize a decision tree classifier with the diabetes training dataset, leveraging a lake house, notebook, and pipeline.

Explore the core ideas of deep learning, including neural networks, layers, weights and biases, activation functions like ReLU, and loss with backpropagation across training epochs in large language models.

Build a deep learning diabetes predictor in a Fabric lake house using PyTorch, notebook workflows, 70/30 data splits, and model inference.

Explore Azure AI services—language, vision, and speech—through cognitive and applied AI offerings, and learn how Fabric notebooks publish these workloads for sentiment analysis and invoice extraction with Azure Document Intelligence.

Develop an end-to-end sentiment analysis workflow for e-commerce platform by integrating Azure Language Services with Microsoft Fabric and building Power BI insights from a GitHub dataset and lakehouse semantic model.

Deploy an Azure Language Service resource in portal.azure.com and explore its pre-built capabilities, including sentiment analysis, key phrase extraction, named entity recognition, and text summarization via Language Studio.

Build a sentiment analysis workflow in fabric using Azure AI Text Analytics, data lake house, and data flows to process e-commerce reviews and visualize results in Power BI.

Explore the evolution from artificial intelligence to generative AI and large language models, and learn how transformer-based foundation models power applications like chatbots and Copilot.

Discover how Azure OpenAI delivers tenant-isolated, secure access to OpenAI models like GPT and text embedding models through private endpoints and the Azure OpenAI Studio for deployment.

Master prompt engineering by crafting precise prompts that define goal, context, expectations, and sources to guide AI agents. Narrow focus yields accurate, meaningful results across language, code, and images.

Launch an Azure OpenAI resource in portal.azure.com, deploy GPT-4, and obtain keys and endpoints to call the model via Azure OpenAI Studio for chat and image tasks.

Learn to call the GPT engine with the chat completions API using the Azure OpenAI SDK in Python, connect to an Azure OpenAI resource, and configure keys and endpoint.

Vector embeddings convert words, phrases, and images into high-dimensional vectors such as Ada 002's 1536 dimensions, capturing semantic meaning to support retrieval augmented generation with Azure OpenAI.

Learn to generate vector embeddings from Python code using an Azure OpenAI text embedding 002 model, configure environment variables, and run the lab to create 1536-dimension embeddings.

Explore how generative AI fits into the ecosystem with Microsoft Fabric and Azure Databricks, enabling a single source of truth from data lake storage to hosted models.

Build a hands-on chatbot lab in fabric notebooks that chats with a GPT model hosted in Azure OpenAI, installing the OpenAI SDK and handling root certificates with certify.

Learn how retrieval augmented generation grounds LLMs in enterprise data with a retrieval pipeline, vector embeddings, and Azure AI Search, including multimodal retrieval augmented generation.

Explore retrieval augmented generation with Microsoft Fabric, Azure OpenAI, and Azure AI Search, detailing vector embeddings, data sources, data flows, and the end-to-end architecture.

Deploy and connect Azure search resource and Azure OpenAI resource to build a Rag architecture using a diabetes FAQ CSV dataset, a Jupyter notebook, and a lake house setup.

Deploy a RAG workflow in Fabric by building a lake house, uploading data, creating embeddings with SynapseML and Azure OpenAI, and indexing them in Azure Search for retrieval augmented generation.