LangGraph Agentic Workflows

A practical guide and reference implementation for building cyclic, stateful AI agent workflows using LangGraph, persistent checkpointers via MongoDB, and local LLMs via Ollama.

🚀 Overview

This repository demonstrates the core building blocks of agentic architectures:

• State Management: Typed schemas defining shared runtime memory across steps.
• Nodes as Functions: Dedicated compute units handling input, generation, and side effects.
• Edges & Dynamic Routing: Conditional branching and LLM-as-a-judge evaluation patterns.
• Persistent Memory & Checkpointing: Persisting conversation state across sessions using MongoDB checkpoints and thread-scoped execution (thread_id).

📁 Project Structure

• docker-compose.yml: Local multi-container setup running MongoDB with persistent volume storage.
• chat_1.py: Basic linear graph with custom state, chatbot node, and messages.
• chat_2.py: Adding multiple sequential nodes and message accumulators.
• chat_3.py: Conditional edges, dynamic routing, and fallback models.
• chat_4_persistent_memory.py: Stateful workflow utilizing MongoDBSaver checkpointer to retain conversation history across independent runs keyed by thread_id.

🛠️ Getting Started

1. Setup Environment

python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate
pip install -r requirements.txt

2. Run Local Models (Ollama)

Ensure your Ollama Docker container or local instance is running:

ollama run qwen2.5:7b
ollama run gemma4:e2b

3.Spin Up MongoDB (For Checkpointing)

Start the local MongoDB service using Docker Compose:

docker compose up -d

4. Run a Workflow

Run linear or conditional routing workflows:

python chat_3.py

Run persistent checkpointed workflows:

chat_4_persistent_memory.py