GenAI and Agentic AI BootCamp

Topics:

• Prompt Structure, CoT, ToT, ReAct,
• Few-Shot Prompting,
• Context Engineering
• Context Windows
• Context Compression
• Context Pruning
• Context Routing
• Context Budgeting
• Managing Large Contexts

Learning Objectives:

Should be able to understand how to write prompts for generating desired output from LLM

Topics:

• Embeddings
• Chunking Strategies
• Vector Databases (FAISS, Pinecone, Weaviate)
• Hybrid Search
• Query Rewriting
• Multi-Query Retrieval
• Multi-Document Retrieval
• LangChain
• RAG Evaluation

Projects:

• Enterprise Knowledge Intelligence

Assistant: Build a RAG system that can search, retrieve, and answer questions from multiple company documents using
FAISS/Pinecone, LangChain, and LlamaIndex.

• Agentic Research Assistant:

Build an agent that decides when to retrieve information, performs multiple retrieval loops, validates results, and generates research reports.

Topics

• Fine-Tuning Fundamentals
• LoRA,
• QLoRA
• Dataset Preparation
• Fine-Tuning vs RAG
• Hugging Face Fine-Tuning Workflow
• GPT-4o Vision
• Claude Vision
• OCR
• PDF Extraction
• Invoice Processing
• Resume Parsing
• Contract Analysis
• Document Intelligence Agents

Projects:

• Document Intelligence Agent:

Upload PDFs, invoices, resumes, and contracts. Extract structured information and answer questions using Vision + OCR + LLMs.

• Custom Domain AI Assistant:

Fine-tune a small model using LoRA/QLoRAon domain-specific data and compare performance against a RAG-based approach.

Topics:

• What is an AI Agent?
• LLM as reasoning engines
• Planner, Executor, Tools, and Memory

Learning Outcomes:

• Introduce the concept and components of AI agents.

Use Cases/Projects:

• Understand the agent architecture and roles of each part.

Learning Objectives:

• Create a rule-based planning agent with memory.

Assessment Methods:

• Short quiz + code implementation.

Topics:

• AutoGPT, BabyAGI, LangGraph overview, Crew AI framework ReAct, Plan-and-Execute, Tree of Thought
• Goal decomposition and task queues

Learning Outcomes:

• Explore agentic frameworks and architecture patterns.

Use Cases/Projects:

• Able to choose and use frameworks like AutoGen and CrewAI.

Learning Objectives:

• Build a simple CrewAI or LangGraph agent pipeline.

Assessment Methods:

• Demo and code walkthrough.

Topics:

• Agentic RAG
• Langgraph
• web search
• tool
• vectorDB

Topics:

• CrewAI Framework
• AutoGen Framework
• LangGraph Framework
• Agent Orchestration Patterns
• Framework Comparison and Selection

Learning Outcomes:

• Learn about different Frameworks

Learning Objectives:

• Able to create use cases for different frameworks

Topics:

• API Calling
• Browser Tools
• File System Operations
• Function Calling
• Structured Outputs
• JSON Mode
• Pydantic Validation
• Output Parsers
• Retry Strategies
• Reliable Agent Responses

Learning Objectives:

• Enable agents to interact with external tools and services.

Learning Outcomes:

• Agents can fetch, read, write, and manipulate data.

Use Cases/Projects:

• Build an agent that queries web APIs and updates documents.

Assessment Methods:

• Code demo and test scenarios.

Topics:

• Short-Term Memory
• Long-Term Memory
• Vector Memory
• FAISS
• Pinecone
• LangGraph Memory Persistence
• Context Retention Strategies
• Learning Objectives:
• Implement agent memory for enhanced context.

Learning Outcomes:

• Create agents with persistent vector memory.

Use Cases/Projects:

• Create a chatbot with persistent user memory.

Assessment Methods:

• Functionality demonstration with memory storage.

Topics:

• Multi-Step Planning
• ReAct Planning
• Dynamic Tool Chaining
• Decision Making Under Constraints
• Goal-Oriented Agents

Learning Objectives:

• Enable agents to reason and plan across multiple steps.

Learning Outcomes:

• Agents can dynamically adapt plans and tool usage.

Use Cases/Projects:

• Build a goal-seeking assistant with dynamic steps.

Assessment Methods:

• Step-by-step task execution assessment.

Topics:

• CrewAI
• AutoGen C
• AMEL
• Role-Based Orchestration
• Agent Negotiation
• Message Passing
• Collaborative Agent Systems

Learning Objectives:

• Introduce agents that collaborate and negotiate.

Learning Outcomes:

• Design multi-agent systems with defined roles.

Use Cases/Projects:

• Create buyer-seller agents that simulate transactions.

Assessment Methods:

• Group demo with inter-agent communication.

Topics:

• Dockerizing and deploying agents
• Hugging Face Spaces
• AWS Lambda
• GCP Deployment
• LLMOps Basics
• Deployment Strategies
• Production Readiness

Learning Objectives:

• Deploy agents in production-ready infrastructure.

Learning Outcomes:

• Deploy scalable and maintainable agent applications.

Use Cases/Projects:

• Deploy a LangGraph agent on HF Spaces with Docker.

Assessment Methods:

• Deployment report and live link submission.

What You’ll Achieve:

• Build AI agents using Azure AI Studio (AI Foundry)
• Deploy production-ready agent systems on cloud infrastructure
• Implement enterprise-grade RAG using Azure AI Search
• Understand how large organizations design AI agent workflows
• Complete a Microsoft Learn Applied Skills badge as part of the program

WhatYou’ll Learn:

• Azure AI Foundations
  Setting up Azure AI Studio and environments
• Understanding Azure OpenAI services Agent Development on Azure
• Building agents using Azure-native tools Orchestrating workflows with Semantic Kernel
• Enterprise RAG Systems
• Connecting agents to business data using Azure AI Search
• Designing knowledge-driven AI systems
• Deployment & Scaling
• Deploying agents on Azure cloud
• Monitoring, logging, and performance optimization
• Cost & Practical Usage
• Using Azure free credits efficiently Avoiding unnecessary API costs

Hands-On:

• Build an enterprise knowledge assistant (RAG-based agent)
• Deploy an AI workflow automation agent on Azure
• Convert your existing agent (CrewAI/LangGraph) into a cloud system
• Complete Microsoft Learn labs with instructor guidance

Microsoft Learn Badge (Applied Skills):

As part of this module:

• You will complete a Microsoft Learn AI Agents course on Azure
• Get step-by-step guidance to finish all required labs
• Use Azure free credits to complete the tasks
• Upon successful completion, you will receive an official Microsoft Learn Applied Skills badge directly from Microsoft

Topics:

• Loop prevention and constraint design
• Goal alignment and ethical agents
  Trust boundaries and permission control
• Agent Governance
• Practical prompt injection defense — input sanitization, output filtering, tool permission scoping, sandboxing agent actions

Learning Objectives:

• Ensure safe, ethical, and aligned agent behaviors.

Learning Outcomes:

• Able to implement safety constraints and ethical checks.

Use Cases/Projects:

• Implement constraints in a goal-chasing agent.

Assessment Methods:

• Checklist-based review + discussion.

Topics:

• Monetizable Agents,
• Travel Planner Agent,
• Sales Bot, Scheduling,
• Persistent Jobs,
• Dashboards,
• SaaS Business Models

Learning Objectives:

• Productize agents and offer as commercial services.

Learning Outcomes:

• Design agent SaaS apps for real-world use cases.

Use Cases/Projects:

• Build and price a travel planning SaaS agent.

Assessment Methods:

• Business model presentation + prototype demo.

Topics:

• Introduction to MCP (Multi-Agent Communication Protocols) Understanding A2A
  (Agent-to-Agent) Protocol
• Deep Dive into ACP (Agent Collaboration Protocol)
• Comparing MCP vs A2A vs ACP
• When and how to apply each protocol in real-world agent systems
• Communication Patterns, Collaboration Workflows

Learning Objectives:

• Understand the core principles and differences between MCP, A2A, and ACP protocols.
• Learn when and how to apply each protocol in real-world multi-agent systems.
• Gain practical exposure to implementing these protocols using frameworks like AutoGen and CrewAI.

Learning Outcomes:

• Ability to design and implement agent communication using MCP, A2A, and ACP.
• Confidently choose the right protocol based on system requirements and agent behavior.
• Deliver functional multi-agent workflows with appropriate communication and collaboration logic.

• n8n overview, architecture, and core concepts
• Workflow creation using triggers, nodes, and executions
• Data handling with JSON, expressions, conditions, and errors
• Integrations with APIs, databases, and third-party tools
• Advanced workflows, AI agents, deployment, and best practices