⚡ AgenticJVM

From public static void main(String[] args) to Autonomous Multi-Agent Systems

A structured, code-first journey through Java — and into the agentic AI stack built on top of it.

---

🧭 What This Is

AgenticJVM is a public, code-backed learning system that takes the JVM ecosystem all the way from first principles to production-grade autonomous agents.

Most "agentic AI" content online lives in Python. This repo is the opposite bet: everything here is built on Java — the language most enterprise systems actually run on — using the frameworks now bringing agentic AI into that world: Spring AI, Google's Agent Development Kit (ADK), LangChain4j, and the Model Context Protocol (MCP).

Why public? Because the best way to prove you understand a stack this new is to build in the open and let the commit history, issues, and shipped agents speak for themselves.

---

🗺️ The Roadmap

graph LR
    A[Java Fundamentals] --> B[Core + Advanced Java]
    B --> C[Spring & Spring Boot]
    C --> D[LLM Foundations]
    D --> E[Spring AI]
    E --> F[Google ADK]
    F --> G[LangChain4j]
    G --> H[MCP]
    H --> I[Shipped Agents]

Loading

Block	Focus	Status
00	Java Fundamentals (syntax → OOP basics)	⬜
01	Core Java (collections, streams, generics, concurrency)	⬜
02	Advanced Java (JVM internals, reflection, design patterns)	⬜
03	Spring Framework (DI, AOP, Spring MVC)	⬜
04	Spring Boot (auto-config, REST, data, testing)	⬜
05	LLM Foundations — NLP → Transformers → LLMs in Java	⬜
06	Spring AI — ChatClient, RAG, Advisors, Vector Stores	⬜
07	Google ADK — LlmAgent, workflow & multi-agent patterns	⬜
08	LangChain4j — AI Services, Supervisor & P2P agentic patterns	⬜
09	MCP — building & consuming Java MCP servers	⬜
10	Capstone Projects	⬜

Tracker is also maintained live as GitHub Issues — one issue per syllabus topic, closed as completed.

---

📂 Repository Structure

agentic-jvm/
├── 00-java-fundamentals/
├── 01-core-java/
├── 02-advanced-java/
├── 03-spring-framework/
├── 04-spring-boot/
├── 05-llm-foundations/
│   └── nlp-to-transformers/
├── 06-spring-ai/
│   ├── chatclient-prompts/
│   ├── memory-advisors/
│   └── rag-pipelines/
├── 07-google-adk/
│   ├── first-agent/
│   ├── workflow-agents/        # sequential, parallel, loop
│   └── multi-agent-a2a/
├── 08-langchain4j/
│   ├── ai-services/
│   ├── tool-calling/
│   └── supervisor-p2p/
├── 09-mcp/
│   ├── mcp-server-java/
│   └── mcp-client-integration/
├── 10-projects/
│   ├── ai-travel-planner-agent/      → Google ADK
│   ├── ai-customer-support-bot/      → LangChain4j
│   └── ecom-ai-capstone/             → Spring AI + PGVector + ADK
├── docs/
│   └── diagrams/
└── LEARNING_LOG.md

Each numbered module has its own README.md with: learning objectives → key concepts → runnable code → what broke and why.

---

🚀 Featured Builds

✈️ AI Travel Planner Agent — Google ADK

Multi-agent system (Flight Agent + Hotel Agent + Itinerary Agent) coordinated by an orchestrator. Takes destination, budget, and dates; returns a compiled, end-to-end travel plan using live flight/hotel tool integrations.

[📁 Code](./10-projects/ai-travel-planner-agent) · [🔗 Live Demo](#)

🎧 AI Customer Support Bot — LangChain4j

RAG-powered support agent grounded in product docs and FAQs, with persistent conversation memory, tool calling for order status/ticketing, and agentic routing to department-specific sub-agents.

[📁 Code](./10-projects/ai-customer-support-bot) · [🔗 Live Demo](#)

🛒 E-Com AI Capstone — Spring AI + PGVector + ADK

Full-stack AI commerce backend: semantic product search via PGVector, a ChatClient-backed conversational assistant, DALL·E-generated product visuals, and ADK agents handling order management — with Grafana/Prometheus observability wired in.

[📁 Code](./10-projects/ecom-ai-capstone) · [🔗 Live Demo](#)

---

🛠️ Tech Stack

Language & Core: Java 21, Maven/Gradle Frameworks: Spring Boot 3.x, Spring AI 2.0, LangChain4j, Google ADK Agentic Infra: Model Context Protocol (MCP), Agent-to-Agent (A2A) protocol Vector & Storage: PGVector, Redis Vector Store, SimpleVectorStore Models: OpenAI, Anthropic, Hugging Face, Ollama (local) Observability: Grafana, Prometheus, Spring AI Evaluators Deployment: Docker, AWS SageMaker, Google Cloud Run

---

🏁 Getting Started

Prerequisites

• JDK 21+
• Maven 3.9+ or Gradle 8+
• Docker (for PGVector / Redis locally)
• An OpenAI/Anthropic API key (or Ollama for local models)

# Clone
git clone https://github.com/YOUR_USERNAME/agentic-jvm.git
cd agentic-jvm

# Run any module independently, e.g.:
cd 06-spring-ai/rag-pipelines
mvn spring-boot:run

Each module's own README.md has module-specific setup (API keys, Docker compose for vector DBs, etc.).

---

📓 Learning Log

Weekly devlogs on what was built, what broke, and what clicked differently the second time through — kept in LEARNING_LOG.md and mirrored in Discussions.

---

📜 License

MIT — see LICENSE. Use it, fork it, learn from it.

---

Built by krish kumar · LinkedIn · Portfolio

If this helped you learn agentic AI on the JVM, a ⭐ helps others find it too.