Agentic AI Engineering

Agentic AI Engineering is a production-grade engineering resource for building modern agentic AI systems with LangChain, LangGraph, RAG, MCP, local models, and deployable Python services.

The repository leads toward the architecture implemented in Chapter 5: a multi-node LangGraph assistant connected to a standalone MCP server, with intent routing, tool execution, response summarization, email tooling, math tools, automated tests, and GitHub Actions CI. Earlier chapters build the required layers underneath it: provider abstraction, LCEL orchestration, vector retrieval, memory, ReAct agents, router graphs, sequential workflows, multi-agent collaboration, and human-in-the-loop control.

This is not a beginner chatbot walkthrough. It is a structured engineering path for developers building systems that need state, tools, routing, retrieval, observability, modularity, and model-provider flexibility.

What You'll Build

• Multi-node LangGraph assistant with router, execution, summarization, and conversation nodes.
• MCP tool server exposing isolated math and email tools over a decoupled server boundary.
• Tool-using ReAct workflows that call external capabilities through typed tool contracts.
• RAG pipelines using vector stores, embeddings, retrieval chains, and local document context.
• Provider-flexible LLM interfaces across OpenAI, Gemini, and Ollama.
• Agent routing systems for sequential, router-based, ReAct, and multi-agent workflows.
• Human-in-the-loop execution paths for safer agent behavior in production-style flows.
• Tested agent components with pytest coverage for Chapter 5 node behavior and tool contracts.
• CI-backed agent validation through GitHub Actions for repeatable checks on graph and MCP behavior.

Engineering Roadmap

System Architecture Focus

Chapter 5 is the reference implementation for the repository's production architecture:

• LangGraph StateGraph coordinates routing, execution, summarization, and conversation paths.
• Router node classifies user intent before allocating work to the right execution path.
• Execution node binds LangChain tool-calling agents to MCP-exposed capabilities.
• Summarization node converts raw tool responses into user-facing output.
• Conversation node handles non-tool interactions without invoking the heavier tool path.
• FastMCP server isolates operational tools from the agent runtime.
• Pytest suite and GitHub Actions CI validate tool behavior, routing assumptions, and graph execution contracts.

Start with the full implementation here: Chapter 5: Multi-Node LangGraph Agent with MCP Tools.

Tech Stack

• Agent frameworks: LangChain, LangGraph
• Protocols: Model Context Protocol (MCP), FastMCP
• LLM providers: OpenAI, Google Gemini, Ollama
• Retrieval: ChromaDB, FAISS, Sentence Transformers
• Backend tooling: Python, FastAPI, pytest, GitHub Actions
• ML ecosystem: PyTorch, TensorFlow, Hugging Face

Repository Structure

.
├── Chapter1/                         # LLM providers, streaming, prompt foundations
├── Chapter2/                         # LangChain LCEL, chains, tools, routing
├── Chapter3/                         # Memory, retrieval, vector stores, RAG
├── Chapter4/                         # LangGraph agent patterns and HITL workflows
├── Chapter5/
│   └── SimpleChatAgent/
│       ├── AganticAssistant/         # LangGraph assistant runtime
│       ├── McpServer/                # MCP tool server
│       ├── tests/                    # pytest coverage for tools and nodes
│       └── demo/                     # screenshots and demo recordings
├── requirements.txt
└── README.md

Quick Start

1. Clone the Repository

git clone [email protected]:zkzkGamal/agentic-ai-engineering.git
cd agentic-ai-engineering

Or with HTTPS:

git clone https://github.com/zkzkGamal/agentic-ai-engineering.git
cd agentic-ai-engineering

2. Create an Environment

python3 -m venv venv
source venv/bin/activate

For Windows:

venv\Scripts\activate

3. Install Dependencies

Install the full repository stack:

pip install -r requirements.txt

Or install only the chapter you are running:

pip install -r Chapter1/requirements.txt
pip install -r Chapter2/requirements.txt
pip install -r Chapter3/requirements.txt
pip install -r Chapter4/requirements.txt
pip install -r Chapter5/SimpleChatAgent/AganticAssistant/requirements.txt
pip install -r Chapter5/SimpleChatAgent/McpServer/requirements.txt

4. Configure Environment Variables

Each chapter that needs credentials includes its own .env.example.

cp Chapter1/.env.example Chapter1/.env
cp Chapter2/.env.example Chapter2/.env
cp Chapter3/.env.example Chapter3/.env
cp Chapter4/.env.example Chapter4/.env

Chapter 5 has separate runtime boundaries for the assistant and MCP server:

cp Chapter5/SimpleChatAgent/AganticAssistant/.env.example Chapter5/SimpleChatAgent/AganticAssistant/.env
cp Chapter5/SimpleChatAgent/McpServer/.env.example Chapter5/SimpleChatAgent/McpServer/.env

Running the Chapter 5 Agent System

Start the MCP server:

python3 Chapter5/SimpleChatAgent/McpServer/main.py

In another terminal, run the LangGraph assistant:

python3 Chapter5/SimpleChatAgent/AganticAssistant/main.py

Run the Chapter 5 test suite:

cd Chapter5/SimpleChatAgent
pytest tests

Chapter Details

Chapter 1: LLM Provider Foundation

Direct integration with OpenAI, Gemini, and Ollama. Covers provider calls, streaming responses, prompt structure, and model-facing interfaces used later by orchestration and agent layers.

Chapter 2: LangChain Orchestration Layer

Builds LCEL pipelines, sequential chains, router chains, prompt templates, output handling, and tool binding. This chapter establishes the composition patterns used by larger agent systems.

Chapter 3: Retrieval and Memory Infrastructure

Implements conversation memory, entity tracking, embeddings, local vector stores, and RAG flows with ChromaDB, FAISS, and Sentence Transformers.

Chapter 4: Agent Graph Patterns

Moves from chains into stateful agent workflows with LangGraph StateGraph. Covers ReAct, router agents, sequential agents, multi-agent collaboration, self-refine loops, and human-in-the-loop control.

Chapter 5: Multi-Node LangGraph and MCP Runtime

Implements a decoupled agent runtime with a LangGraph assistant and MCP server. The assistant routes requests across specialized nodes, invokes MCP tools, summarizes tool outputs, and validates behavior with automated tests and CI.

Production Repositories

Author

Zkaria Gamal - AI Engineer

• GitHub: @zkzkGamal
• LinkedIn: Zkaria Gamal

License

This repository is available under the MIT License.