[Link].

com
 Fundamentals:
NLP to LLM
Architecture

Spring Al:
RAG & Memory

Google ADK:
Workflow
Agents

LangChain4j:
Supervisor &
P2P

Build:
Travel Agent
Support Bot
E-Com AI

[Link]
 01
Step 01: Lay the
Neural Foundation
NLP Fundamentals Deep Learning
What is NLP and why it AI / ML / DL hierarchy
matters for AI engineers explained
Core NLP tasks: Tokenization, Neural network fundamentals:
POS Tagging, NER weights, layers, activation
Sentiment analysis and text Attention mechanism deep
classification dive
Word2Vec and GloVe: Why deep learning replaced
pre-transformer embeddings classical NLP approaches
From classical NLP to Deep
Learning to LLMs

Transformers LLMs in Java

Transformer architecture: API vs SDK vs Framework:
encoder and decoder which to pick
BERT vs GPT: bidirectional vs OpenAI REST API calls directly
autoregressive from Java
Multi-head attention and Official OpenAI Java SDK
positional encoding walkthrough
LLMs: Types, Tokens and Hugging Face integration with
Limitations Spring AI
Hallucination: why it happens Building your first AI-powered
and how to reduce it Java app

[Link]
Step 02: Spring AI

[Link]
[Link]
 03
Step 03: Google ADK

[Link]
Orchestrate Multi-Agent
Workflows

[Link]
 04
Step 04: LangChain4j

[Link]
[Link]
[Link]
 Step 05: Build 05
Autonomous Systems

[Link]
The Capstone: Al-Powered E-Com App

Frontend UI
(React)

[Link]
[Link]
NAVIN REDDY HYDER ABBAS

[Link]
Mastering Agentic AI with Java
Date: June 14, 2026 Format: Weekend (Sat & Sun, 9 AM - 12 PM IST)
Duration: 4 Months Prerequisites: Java & Spring Boot

Syllabus Overview

This course is engineered for Java developers transitioning from basic LLM integrations
to building intelligent, autonomous AI systems. Theory is minimized; every concept is tied
directly to production-ready Java code.

Core Outcomes & Architecture:

• Frameworks: Mastery of Spring AI 2.0, LangChain4j, and Google's Agent Development
Kit (ADK).
• Agentic Patterns: Implement sequential, parallel, loop, conditional, supervisor, and peer-
to-peer multi-agent systems.
• Core Mechanisms: Deep dive into prompt engineering (ReAct, CoT), Model Context
Protocol (MCP), and tool calling.
• Infrastructure: Integrate Vector Databases (PGVector, Redis, ChromaDB) and construct
advanced RAG pipelines.
• Production Deployment: Implement multimodal AI, model evaluation, fine-tuning on
AWS SageMaker, and observability using Grafana and Prometheus.
• Projects : Build an AI Travel Planner Agent using Google ADK, an AI Customer
Support Bot using LangChain4j, and an AI-powered e-commerce backend featuring smart
product search, conversational assistant, order management agents, and full monitoring.

[Link]
Detailed Syllabus

Block 1: Foundations
What is Agentic AI
What is Agentic AI and why it matters right now
Autonomous agents vs traditional LLM chat applications
Key properties of agents: reasoning, planning, memory, and tool use
Agentic AI in the real world: industry use cases and adoption
Frameworks we will cover: Spring AI, Google ADK, LangChain4j, and MCP

NLP FUNDAMENTALS
What is NLP, and why does it matter for AI engineers
Core NLP tasks: Tokenization, POS Tagging, NER
Sentiment analysis and text classification
Word2Vec and GloVe: pre-transformer embeddings
From classical NLP to Deep Learning to LLMs (the bridge)

AI, ML, Deep Learning & Transformers

AI / ML / DL hierarchy
Neural network fundamentals: weights, layers, and activation
Attention mechanism deep dive
Transformer architecture: encoder and decoder
BERT vs GPT and their use cases

LLMs: Types, Tokens & Limitations

Types of LLMs: open source vs closed source
Tokens, context windows, and output generation
Temperature, top-p, and sampling strategies
Hallucination: why it happens and how to reduce it
Cost, latency, and rate limits in production

Java + LLMs: First Code

API vs SDK vs Framework: which to pick
OpenAI REST API calls directly from Java
Official OpenAI Java SDK walkthrough
Exploring other Java AI SDKs
Building your first AI-powered Java app

Hugging Face Integration

What is Hugging Face
HF Inference API: setup and authentication
Calling HF API from Java (REST)
Spring AI + Hugging Face provider

[Link]
Detailed Syllabus

Block 2: Spring AI
Spring AI Introduction
Why Spring AI for Java developers
Spring AI 2.0 architecture overview
Model providers overview: OpenAI, Anthropic, Azure, Ollama, Hugging Face
Spring AI auto-configuration and starters
Spring AI use cases for backend Java developers
Spring AI docs walkthrough
Creating a Spring AI project from scratch

ChatClient & Prompts

Create OpenAI API key and configure Spring AI
ChatModel vs ChatClient: understanding the difference
Message types: SystemMessage, UserMessage, AssistantMessage
ChatClient fluent API deep dive
ChatResponse, Generation, and Metadata
[Link] pattern and default system prompt configuration
PromptTemplate and template variables
StringTemplate (.st) files for complex prompt management
Multi-turn conversations with ChatClient

Memory & Advisors

Spring AI Advisors framework explained
MessageChatMemoryAdvisor and PromptChatMemoryAdvisor
InMemoryChatMemoryRepository • JdbcChatMemoryAdvisor with database persistence
QuestionAnswerAdvisor for RAG integration
SafeGuardAdvisor for content moderation
Built-in Advisors overview
Building Custom Advisors
Advisor ordering and chaining

Open Source Models + Prompt Engineering

Running models locally with Ollama
Spring AI with Ollama configuration
Prompt Template basics and use cases
Few-shot prompting
Chain-of-Thought (CoT) reasoning
ReAct prompting pattern
System Prompt vs User Prompt

[Link]
Detailed Syllabus

Vector Embeddings
What are embeddings and vector spaces
Embedding via OpenAI API client
Embedding using Spring AI EmbeddingModel
Comparing embedding models
Visualizing similarity with cosine distance

Vector Databases
Cosine similarity: theory and Java implementation
Vector DB introduction and options
SimpleVectorStore for local dev and testing
Token Text Splitter strategies
Metadata and filtering in vector stores

PGVector & Redis Vector Store

PGVectorStore introduction and setup
PGVector with Docker
PGVector implementation in Spring AI
Redis Vector Store configuration
Redis Vector Store queries and search

RAG: Basics
What is RAG, and why does it work
Basic RAG implementation in Spring AI
DocumentReaders: PDF, Web, JSON, and Markdown loaders
DocumentTransformers pipeline
RetrievalAugmentationAdvisor in Spring AI 2.0

Advanced RAG
Hybrid search: keyword + semantic combined
Metadata filtering strategies
Reranking post-retrieval results
VectorStoreDocumentRetriever and configuration
Semantic caching for cost reduction

[Link]
Detailed Syllabus

Output Converters
Structured Output Converter
List Output Converter
Bean Output Converter
Bean Output Converter with List
When to use each converter pattern

Tool Calling
Tool calling concepts and the ReAct pattern
Tool calling in Spring AI: basics
Defining and registering tools with @Tool
Chained tool calls and multi-step reasoning

MCP: Model Context Protocol

MCP overview and architecture
MCP in Spring AI setup
Building an MCP server in Java
MCP client integration
MCP + Tool Calling working together

Multimodality: Images
OpenAI image model (DALL-E 3) overview
ImagePrompt and ImageResponse
Image generation options and parameters
Describe image feature: vision input
Implementing image understanding in Spring AI

Multimodality: Audio
Audio models overview: STT and TTS
Audio transcription (Whisper): part 1
Audio transcription: part 2 and edge cases
Transcription options and language tuning
TTS speech model and voice options

[Link]
Detailed Syllabus

Model Evaluation
Why model evaluation matters
Spring AI Evaluator API overview
RelevancyEvaluator: checking answer relevance
FactCheckingEvaluator: grounding responses
Custom Evaluators with Spring AI
Evaluation metrics: faithfulness, precision, and recall
Automated evaluation pipeline setup

Fine-tuning & Cloud Model Hosting

What is Fine-tuning
When to Fine-tune vs RAG vs Prompt Engineering
Fine-tuning with OpenAI API (demo)
AWS SageMaker overview for AI engineers
Fine-tuning and deploying a model on SageMaker
Calling SageMaker endpoints from Spring AI

Streaming, Observability & Production

Streaming responses in Spring AI
Prompt caching for cost optimization
Guardrails and content moderation
Spring AI observability integration
Rate limiting, retry, and error handling patterns

Monitoring with Grafana & Prometheus

Prometheus metrics setup in Spring Boot
Spring AI metrics and tracing
Grafana dashboard for AI app monitoring
Production deployment best practices
Performance tuning for LLM-backed apps

[Link]
Detailed Syllabus

Block 3: Google ADK

Google ADK: Introduction & Architecture

What is Google ADK and why use it for Java
ADK architecture overview: Agents, Tools, Sessions, Memory, Runner
ADK for Java vs ADK for Python: key differences
Setting up Google ADK in a Java project
ADK Web UI, CLI, and API Server: four ways to run agents

Google ADK: Building Your First Agent

LlmAgent: structure and configuration
Writing agent instructions effectively
Model selection for ADK: Gemini, Claude, Ollama
Agent Config: temperature, tokens, and safety settings
Running and testing your first ADK agent

Google ADK: Tools & Function Calling

Function tools in ADK: basics and setup
Adding tools to LlmAgent
OpenAPI tools integration
Tool authentication and limitations
MCP tools with ADK: McpToolset class

Google ADK: Sessions, State & Memory

Runner, Session, and State explained
InMemory Session Service
Persistent session management
Adding memory to ADK agents
Context management and context compression in ADK

[Link]
Detailed Syllabus

Google ADK: Workflow Agents

Sequential agents: chaining tasks in order
Parallel agents: concurrent execution
Loop agents: iterative task refinement
Custom agents: building your own agent logic
Agent routing and conditional flows
Human-in-the-Loop (HITL) Workflows

Google ADK: Multi-Agent Systems

Multi-agent architecture overview
Orchestrator and sub-agent patterns
Agent-to-Agent (A2A) communication protocol
A2A in Java: exposing and consuming agents
Debugging and tracing multi-agent flows

Google ADK: Callbacks, Artifacts & Guardrails

Callbacks: before and after model and tool execution
Callback design patterns and best practices
Artifacts: managing files and binary outputs
Implement guardrails and safety settings
Grounding with Google Search and Vertex AI

Google ADK: Observability & Deployment

Observability: logging, metrics, and traces
Evaluating ADK agents: criteria and custom metrics
Deploying agents to Cloud Run using ADK command
Deploying agents to GKE with Docker
Production best practices for ADK agents

[Link]
Detailed Syllabus

Block 4: LangChain4j

LangChain4j: Introduction
What is LangChain4j, and why use it
LangChain4j vs Spring AI: when to pick which
Setting up LangChain4j in a Spring Boot project
Chat and Language Models low-level API
Model parameters: temperature, top-p, timeouts
First chatbot with LangChain4j

LangChain4j: AI Services, Memory & Streaming

AI Services and declarative interface with @AiService
Chat Memory: MessageWindowChatMemory and TokenWindowChatMemory
ConversationMemory and memory providers
Streaming responses with TokenStream
Structured Outputs in LangChain4j
Text Classification with AI Services
Guardrails: input and output validation
Logging and Observability in LangChain4j

LangChain4j: Tool Calling

Tool calling concepts in LangChain4j
@Tool and @P annotations
Chained tool calls and multi-step reasoning
Tool calling with AI Services
Dynamic tool providers

LangChain4j: MCP
MCP overview in LangChain4j context
MCP client setup in LangChain4j
Building a Java MCP stdio server
Connecting MCP server to AI Services
MCP + Tool Calling working together

[Link]
Detailed Syllabus

LangChain4j: Agentic Workflows

The langchain4j-agentic module overview
@Agent annotation and AgenticServices
AgenticScope: shared state across agents
Sequential workflow: chaining agents in order
Loop workflow: iterative refinement with exit conditions
Parallel workflow: running agents simultaneously
Parallel Mapper workflow: fan-out over collections
Conditional workflow: routing based on LLM classification
Optional agents and async agents
Streaming agents in agentic workflows
Error handling and recovery in agentic systems
AgentListener and AgentMonitor for observability
Human-in-the-loop agents
Non-AI agents inside agentic systems
Declarative API for defining workflows with annotations
LangChain4j: Supervisor & Pure Agentic AI
Pure agentic AI vs deterministic workflows
SupervisorAgent: autonomous planning and execution
AgentInvocation planning and response strategies
Supervisor context and context engineering
Goal-Oriented agentic pattern (GOAP)
Peer-to-Peer (P2P) agentic pattern
Building custom Planner implementations
Memory and context sharing across agents in a system
AgentMonitor HTML report generation
Strongly typed inputs and outputs with TypedKey
LangChain4j: RAG & Spring Boot Integration
RAG with LangChain4j EmbeddingStore
Document loaders in LangChain4j
LangChain4j + PGVector / ChromaDB
Integrating LangChain4j into Spring Boot app
Testing and Evaluation of LangChain4j AI apps
LangChain4j vs Spring AI RAG: side-by-side comparison

[Link]
Detailed Syllabus

Block 5: Projects
Mini Project: AI Travel Planner Agent (Google ADK)
Build a travel planning agent using Google ADK
Agent takes destination, budget, and dates as input
Uses tools to search flights, hotels, and attractions
Multi-agent: one for flights, one for hotels, one for itinerary
Final agent compiles and returns a complete travel plan

Mini Project: AI Customer Support Bot (LangChain4j)

Build a customer support chatbot using LangChain4j
RAG-powered: answers from product documents and FAQs
Conversation memory to maintain context across messages
Tool calling to check order status and raise tickets
Agentic workflow: routes queries to the right department agent

Capstone: E-com AI Features

Project intro: E-com AI architecture overview
UI with AI integration
ChatClient-backed product assistant
AI Image Generator (DALL-E) for products
Ask AI feature: conversational product help

Capstone: RAG, Agents & Monitoring

Setting up PGVector in the project
Embedding product and order data
Product Smart Search with semantic vector queries
Advanced RAG upgrade in the project
Tool Calling + ADK agent for order management
Full monitoring setup and production deployment

Step Into the Course

Sign Up Online Pick a Free Trial Enroll Now! Start Learning!

JOIN NOW!
[Link]