5-Day AI Agents Intensive Course with Google

Welcome! This repository contains my personal notes, code, and solutions for the 5-Day AI Agents Intensive Course hosted by Google.

This 5-day online program, crafted by Google’s ML researchers and engineers, helps developers explore the foundations and practical applications of AI agents.

Course Overview

The curriculum is designed to move from foundational concepts to building production-ready systems. Each day blends conceptual deep dives, hands-on codelabs, and live discussions.

Key concepts covered:

• Core agent components: Models, Tools, Orchestration, Memory, and Evaluation.
• Agent Ops: Reliability, governance, and security.
• Practical application: Building agents with the Agent Development Kit (ADK) and Gemini.
• Interoperability: Using the Model Context Protocol (MCP) for complex, long-running operations.

Technologies & Tools

• Core Model: Google Gemini
• Framework: Agent Development Kit (ADK)
• Protocol: Model Context Protocol (MCP)
• Platform: Kaggle (for Codelabs)
• Language: Python
• Support Tool: NotebookLM

Daily Assignments & Curriculum

This section tracks the daily assignments and materials from the course.

Day 1: Introduction to Agents

• Topic: Introduces a taxonomy of agent capabilities, the need for an "Agent Ops" discipline, and the importance of interoperability and security.
• Whitepaper: Introduction to Agents
• Podcast: Unit 1 Summary (via NotebookLM)
• Codelabs (Kaggle):
  1. Build your first agent using Gemini and ADK
  2. Build your first multi-agent systems using ADK
• Resources:
  • Codelab Troubleshooting Guide
  • Note: Kaggle account phone verification is required for codelabs.

Day 2: Agent Tools & Interoperability (MCP)

• Topic: Focuses on external tools and functions that allow an agent to perform actions or retrieve real-time data. Introduces the Model Context Protocol (MCP) for complex operations.
• Whitepaper: Agent Tools & Interoperability with Model Context Protocol (MCP)
• Podcast: Unit 2 Summary (via NotebookLM)
• Codelabs (Kaggle):
  1. Explore new ways to add tools to extend what your agents can do
  2. Explore best practices for tools, including MCP and long-running operations

Day 3: Context Engineering: Sessions & Memory

• Topic: Focuses on context engineering, the practice of dynamically assembling and managing information in an agent's context window. Defines Sessions (immediate history) and Memory (long-term persistence) to create stateful, personalized AI experiences.
• Whitepaper: Context Engineering: Sessions & Memory
• Podcast: Unit 3 Summary (via NotebookLM)
• Codelabs (Kaggle):
  1. Implement Sessions to manage immediate context in your agents
  2. Implement Memory to create long-term, personalized experiences

Day 4: Agent Quality

• Topics: Focuses on agent quality assurance using a holistic evaluation framework. Covers the technical foundation of Observability (Logs, Traces, Metrics) and scalable feedback loops like LLM-as-a-Judge and Human-in-the-Loop (HITL).
• Whitepaper: Agent Quality
• Podcast: Unit 4 Summary (via NotebookLM)
• Codelabs (Kaggle):
  1. Implement observability to help you debug your agents.
  2. Evaluate your agents.

Day 5: Prototype to Production

• Topics: Focuses on the operational lifecycle of AI agents (deployment, scaling, productionization) and transitioning prototypes to enterprise-grade solutions. Covers multi-agent systems using the Agent2Agent (A2A) Protocol and deploying agents to Vertex AI Agent Engine.
• Whitepaper: Prototype to Production
• Podcast: Unit 5 Summary (via NotebookLM)
• Codelabs (Kaggle):
  1. Explore how to use A2A Protocol to have agents interact with each other
  2. [Optional] Deploy your agent to Agent Engine on Google Cloud.

Acknowledgement & Disclaimer

This repository is for personal learning and educational purposes only.

All course materials, including whitepapers, codelabs, podcasts, and any other associated content, are the intellectual property of Google and Kaggle. All rights, licenses, and acknowledgements are held by them. This repository does not claim any ownership of the original course content.