ADK from Zero — Build Agentic AI with Google ADK

A practical, project-based resource for developers who want to understand Google Agent Development Kit through runnable implementations.

Learn the concept → inspect the implementation → run it → modify it → build your own system.

This repository is my hands-on journey of learning Google's Agent Development Kit (ADK) — a framework for building LLM-powered agents and agentic AI applications.

The examples progress from basic agents and tools to state management, multi-agent systems, and workflow-based architectures.

---

🚀 Features

• Learn Google ADK from beginner to advanced concepts
• Hands-on examples with runnable code
• One shared virtual environment for all examples
• Progressive learning roadmap
• Function and tool calling
• Sessions and state management
• Persistent storage
• Multi-agent architectures
• Sequential, parallel, and loop workflows
• Structured outputs with Pydantic
• Callbacks and agent lifecycle concepts
• Links to official ADK documentation

---

📁 Repository Structure

AgenticAI/
│
├── 01_basic_agent/
├── 02_tool_agent/
├── 03_litellm_agent/
├── 04_structured_outputs/
├── 05_sessions_state/
├── 06_persistent_storage/
├── 07_multi_agent/
├── 08_stateful_multi_agent/
├── 09_callbacks/
├── 10_sequential_agent/
├── 11_parallel_agent/
├── 12_loop_agent/
├── agent-development.md/
├── llm_complete_ceveloper_notes.md/
├── ai_fundamentals.md/
│
├── .gitignore
├── requirements.txt
└── README.md

---

🛠️ Prerequisites

Before starting, make sure you have:

• Python 3.10+
• Git
• A Google AI Studio API key
• Basic Python knowledge

---

⚡ Quick Start

1. Clone the Repository

git clone https://github.com/harsh-chauhan-dev/agentic_ai.git
cd adk-from-zero

2. Create a Virtual Environment

python -m venv .venv

3. Activate the Environment

Windows PowerShell

.venv\Scripts\Activate.ps1

Windows CMD

.venv\Scripts\activate.bat

macOS / Linux

source .venv/bin/activate

4. Install Dependencies

pip install -r requirements.txt

The same virtual environment is used for all examples in this repository.

---

🔑 Setting Up API Keys

The examples use Google's Gemini models through ADK.

1. Create a Google Cloud Project

Create a project through Google Cloud.

2. Create an API Key

Create your API key through Google AI Studio.

3. Configure the API Key

Each example contains a .env.example file.

Navigate to the example you want to run:

cd 01_basic_agent

Rename:

.env.example

to:

.env

Then add your API key:

GOOGLE_API_KEY=your_api_key_here

Security: Never commit your .env file or API keys to GitHub.

If the APIs or services you use require billing, make sure billing is enabled for the associated Google Cloud project.

---

📚 Examples

The repository follows a progressive learning path.

#	Example	What You'll Learn
01	Basic Agent	Create and run a simple ADK agent
02	Tool Agent	Give agents capabilities through tools
03	LiteLLM Agent	Switch between different LLM providers/models
04	Structured Outputs	Use Pydantic and output_schema
05	Sessions & State	Maintain information across interactions
06	Persistent Storage	Persist agent data across sessions
07	Multi-Agent	Orchestrate multiple specialized agents
08	Stateful Multi-Agent	Manage state across complex agent interactions
09	Callbacks	Monitor and respond to agent lifecycle events
10	Sequential Agent	Build sequential agent pipelines
11	Parallel Agent	Execute agent tasks concurrently
12	Loop Agent	Build iterative feedback and refinement workflows

---

🧠 Learning Philosophy

This repository is my hands-on journey of learning Google ADK and Agentic AI by building each concept from scratch.

The goal is not just to copy examples, but to understand:

LLM
 ↓
Agent
 ↓
Tools
 ↓
State
 ↓
Memory
 ↓
Multi-Agent Systems
 ↓
Workflows
 ↓
Real-World Agentic Applications

Each folder focuses on one concept before moving to a more advanced architecture.

---

📖 Official Documentation

For deeper information and the latest features, refer to the official Google ADK documentation:

• Google ADK Documentation
• ADK Quickstart
• Google AI Studio

---

👨‍💻 Author

Harsh Chauhan

• Learning Journey: Agentic AI with Google ADK
• Started: 2026

Built while learning, experimenting, and understanding Agentic AI from the ground up.

If this repository helps you, consider giving it a ⭐.

---

Learning in public. Building step by step. 🚀