Skip to content

Latest commit

 

History

History
165 lines (135 loc) · 10.3 KB

CURRICULUM.md

File metadata and controls

165 lines (135 loc) · 10.3 KB

Azure AI Engineering in 5 Weeks - Curriculum 🚀

Welcome to the Azure AI Engineering in 5 Weeks curriculum. This guide will help you navigate through the weekly lessons, labs, and quizzes designed to enhance your skills in Azure AI services.

Table of Contents 📅

Week by Week 📅

Week 1: Introduction to Azure AI Services 📘

Topics

  • Overview of Azure AI Services:
    • Azure AI Foundry: Learn how to use the most comprehensive AI platform.
    • Azure OpenAI Service: Leverage large language models for NLP tasks.
    • Azure AI Search: Implement AI-powered search within applications.
    • Azure AI Document Intelligence: Process and extract information from documents.
    • Azure AI Vision Services: Image and video analysis tools.
    • Azure AI Speech Services: Speech recognition and synthesis functionalities.
  • Introduction to Retrieval-Augmented Generation (RAG) Patterns:
    • Combining retrieval-based methods with generative AI for enhanced performance.
    • Implementing basic RAG architectures using Azure services.

Practical Session

  • Building Your First Chatbot:
    • Configure necessary Azure services.
    • Develop a chatbot using Azure AI Search and Azure OpenAI through the Azure AI Foundry portal’s UI.

Homework

  • Lab: Complete the lab.ipynb notebook to practice building and refining your first chatbot.
  • Quiz: Test your understanding by filling out the quiz.md file for Week 1.
  • Reinforcement:
    • Upload sample data (Contoso HR documents) and index it using Azure AI Search.
    • Implement hybrid search, combining vector-based and keyword-based retrieval.
    • Enable question-answering capabilities using Azure OpenAI as the reasoning engine.

Note: For setup and detailed instructions, refer to the Week 1 README.

Week 2: Designing Intelligent Cloud-Native AI Systems 🏗️

Topics

  • Introduction to Orchestration Frameworks:
    • Semantic Kernel: Explore Microsoft's open-source framework that facilitates the integration of large language models (LLMs) with conventional programming languages, enabling the creation and orchestration of complex AI solutions.
  • Enhancing Chatbot Functionality with Multimodality:
    • Multimodal AI Integration: Learn how to extend your chatbot's capabilities by incorporating multiple data types, such as text, images, and documents, to provide richer and more context-aware interactions.
  • Vector Database in Azure:
    • Azure AI Search: Dive into building a vector-based data store using Azure AI Search, learning how to efficiently index and search data with integrated vectorization.

Practical Session

  • Upgrading Your Chatbot with Semantic Kernel:
    • Integration: Rebuild the chatbot developed in Week 1 by incorporating Semantic Kernel as the orchestrator, allowing for seamless interaction between various AI services and data modalities.
    • Multimodal Capabilities: Enhance the chatbot to process and respond to inputs beyond text, such as images and documents, utilizing Azure AI Services.
    • Vector Search: Integrate a vector database for storing and retrieving embeddings, enabling efficient multimodal search capabilities.
  • Building a Front-End with Streamlit:
    • User Interaction: Develop a streamlined interface using Streamlit that communicates with the backend, showcasing the chatbot's enhanced multimodal capabilities.

Homework

  • Lab: Work through the lab.ipynb notebook to integrate multimodal features and vector search into your chatbot.
  • Quiz: Complete the Week 2 quiz.md to assess key takeaways and ensure understanding of newly introduced concepts.
  • Reinforcement:
    • Extend your chatbot to handle at least two data modalities (e.g., image and document) with Semantic Kernel and Azure AI Services.
    • Build and query a vector database in Azure AI Search for storing embeddings and performing efficient searches.
    • Advanced (Optional): Deploy the chatbot using the Semantic Kernel framework and Streamlit, containerized with Azure App Services or Azure Container Apps.

Note: For setup and detailed instructions, refer to the Week 2 README.

Week 3: Operationalizing and Managing AI Solutions ⚙️

Topics

  • Transition from MLOps to LLMOps:
    • Understanding the lifecycle of large language models in production.
    • Implementing best practices for deployment and maintenance.
  • Monitoring and Management:
    • Utilizing Azure Monitor, Log Analytics, and KQL for system observability.
    • Setting up end-to-end monitoring with OpenTelemetry and correlation IDs.
  • Continuous Evaluation and Benchmarking:
    • Establishing frameworks for ongoing model assessment.
    • Using Azure AI Studio for evaluation and tracing.

Practical Session

  • Implementing LLMOps Practices:
    • Set up CI/CD pipelines with GitHub Actions for automated deployments.
    • Configure monitoring tools to track system performance.
    • Implement data drift detection and automated retraining workflows.

Homework

  • Lab: Follow the lab.ipynb notebook to add monitoring and management capabilities to your AI system.
  • Quiz: Answer the Week 3 quiz in quiz.md, focusing on operational best practices for AI systems.
  • Reinforcement:
    • Establish comprehensive monitoring and evaluation for your AI solution.
    • Implement an enterprise-grade evaluation framework within your RAG pattern.
    • Optional: Integrate multimodal evaluation to monitor diverse data types effectively.

Note: For setup and detailed instructions, refer to the Week 3 README.

Week 4: Advanced Optimization and Fine-Tuning of AI Systems 🔧

Topics

  • Designing Scalable AI Systems:
    • Architectural patterns for scalability in AI solutions.
    • Leveraging Azure's infrastructure to support large-scale AI deployments.
  • Leveraging Small Language Models:
    • Introduction to small language models and their applications.
    • Understanding the Phi-4 architecture and its benefits.
  • Fine-Tuning Large Language Models (LLMs):
    • Techniques for customizing LLMs for specific domain tasks.
    • Tools and resources available in Azure for model fine-tuning.
  • Enhancing RAG Pattern Performance:
    • Methods to reduce latency and improve accuracy in RAG implementations.
    • Case studies demonstrating performance optimization.

Practical Session

  • Fine-Tuning an Azure OpenAI Model:
    • Hands-on experience in customizing an OpenAI model within Azure.
    • Evaluating performance improvements post fine-tuning.

Homework

  • Lab: Use the lab.ipynb notebook to fine-tune an Azure OpenAI model and analyze the performance improvements.
  • Quiz: Check your knowledge by completing the Week 4 quiz.md, focusing on optimization strategies.
  • Reinforcement:
    • Submit a detailed analysis of your results, highlighting performance gains.
    • Optional: Compare the fine-tuned model with a baseline to assess the impact of your optimizations.

Note: For setup and detailed instructions, refer to the Week 4 README.

Week 5: Building Autonomous AI Systems (Agents) 🤖

Topics

  • Understanding Agent Architectures:

    • Single-Agent Systems: Explore the structure and functionality of standalone AI agents, including their decision-making processes and applications.
    • Multi-Agent Systems: Delve into systems where multiple AI agents interact, collaborate, or compete to achieve individual or collective goals.

  • Designing Multi-Agent System Architectures on Azure:

    • System Design: Learn to architect multi-agent applications using Azure services, ensuring scalability, security, and efficiency.
    • Communication Protocols: Understand the methods and protocols that facilitate effective communication between agents.
    • Coordination Strategies: Study strategies for coordinating tasks among agents to optimize performance and resource utilization.

  • Quickstart: Building Agents in Azure:

    • Azure AI Agent Service: Introduction to Azure's fully managed service for building, orchestrating, and scaling AI agents.
    • Development Tools: Overview of tools like Azure AI Foundry and Azure AI Studio for agent development.
    • Deployment Best Practices: Guidelines for deploying AI agents in production environments, focusing on reliability and performance.

Practical Session

  • Prototyping a Multi-Agent System:
    • Integration with RAG Patterns: Implement a multi-agent system that utilizes Retrieval-Augmented Generation (RAG) to enhance information retrieval and response generation.
    • Agent Collaboration: Set up agents with distinct roles that collaborate to complete complex tasks, demonstrating effective inter-agent communication.
    • Azure Services Utilization: Leverage Azure AI Agent Service and other relevant Azure services to build and deploy your multi-agent system.

Homework

  • Lab: Work through the lab.ipynb to develop a multi-agent system that integrates RAG patterns and demonstrates agent collaboration.
  • Quiz: Update your progress by filling out the Week 5 quiz.md, focusing on agent-based architectures and their practical applications.
  • Reinforcement:
    • Provide comprehensive documentation detailing the system architecture, agent roles, communication protocols, and deployment process.
    • Prepare a demo that illustrates the functionality and effectiveness of your multi-agent system.
    • Optional: Investigate and implement advanced features such as additional data modalities or enhanced agent learning capabilities.

Note: For setup and detailed instructions, refer to the Week 5 README.