🕸️ LangGraph Learning Journey

A structured, notebook-driven deep dive into LangGraph — from basic graph concepts to stateful agents, parallel workflows, human-in-the-loop, subgraphs, and fault-tolerant systems.

📚 What This Repo Is

This repository is my personal learning lab for LangGraph — a low-level framework for building stateful, graph-based AI agents. Each notebook isolates one core concept and pairs it with a real, relatable use case (cricket stats, UPSC essays, X posts, review responses, and more).

The progression covers everything from StateGraph basics through to production patterns like Human-in-the-Loop, subgraphs, persistence, and fault tolerance.

Installation → Simple Graphs → LLM Nodes → Chaining → Parallel → Conditional
      → Iterative Loops → Chatbot → Persistence → HITL → Subgraphs → Fault Tolerance

🗂️ File Structure

langgraph/
│
├── 📓 Notebooks (Learning Path)
│   ├── 0_test_installation.ipynb           # Sanity check — verify setup
│   ├── 1_bmi_workflow.ipynb                # Intro to StateGraph (no LLM)
│   ├── 2_simple_llm_workflow.ipynb         # First LLM node in a graph
│   ├── 3_prompt_chaining_workflow.ipynb    # Sequential prompt chaining as nodes
│   ├── 4_batsman_parallel_workflow.ipynb   # Parallel nodes (fan-out / fan-in)
│   ├── 5_upsc_essay_workflow.ipynb         # Multi-step essay generation
│   ├── 6_quad_eq_conditional_workflow.ipynb      # Conditional edges (router pattern)
│   ├── 7_review_response_conditional_workflow.ipynb  # Sentiment-based routing
│   ├── 8_x_post_generator_iterative_workflow.ipynb  # Iterative loop with refinement
│   ├── 9_basic_chatbot.ipynb               # Basic LangGraph chatbot
│   ├── 10_persistence.ipynb                # State checkpointing & resume
│   ├── 14_hitl.ipynb                       # Human-in-the-Loop interrupts
│   ├── 15_subgraphs.ipynb                  # Nested subgraph composition
│   ├── 15_subgraph_shared.ipynb            # Subgraphs with shared state
│   └── fault_tolerance.ipynb               # Error handling & resilient workflows
│
├── 🐍 Scripts
│   ├── chatbot_with_hitl.py                # HITL chatbot (runnable script)
│   └── chatbot_without_hitl.py             # Baseline chatbot (runnable script)
│
└── .gitignore

🧭 Notebook Breakdown

0️⃣ `0_test_installation.ipynb` — Installation Check

Verifies that LangGraph, LangChain, and OpenAI are correctly installed and your .env is loaded. Run this first before anything else.

1️⃣ `1_bmi_workflow.ipynb` — BMI Workflow (Intro to StateGraph)

No LLM involved — this is a pure graph to teach the mental model. Takes height/weight as input, calculates BMI, routes to a category node.

Concepts: StateGraph, TypedDict state, add_node, add_edge, START, END, .compile(), .invoke()

2️⃣ `2_simple_llm_workflow.ipynb` — Simple LLM Workflow

Adds an LLM node into a graph for the first time. A minimal graph where one node calls ChatOpenAI and returns the result.

Concepts: LLM node functions, MessagesState, binding a model inside a node

3️⃣ `3_prompt_chaining_workflow.ipynb` — Prompt Chaining Workflow

Chains multiple LLM calls as sequential graph nodes — each node's output feeds as context into the next prompt.

Concepts: Sequential edges, passing state between nodes, multi-step reasoning in graphs

4️⃣ `4_batsman_parallel_workflow.ipynb` — Batsman Parallel Workflow

Uses a cricket batsman's stats as input, fans out to multiple parallel analysis nodes (batting avg, strike rate, consistency), then merges results.

Concepts: Parallel node execution, fan-out / fan-in pattern, Send API, merging parallel outputs

5️⃣ `5_upsc_essay_workflow.ipynb` — UPSC Essay Workflow

Generates a structured essay for UPSC (Indian civil services exam) topics using a multi-step pipeline: outline → draft → review → final.

Concepts: Multi-stage LLM pipelines, structured state with multiple fields, prompt engineering within nodes

6️⃣ `6_quad_eq_conditional_workflow.ipynb` — Quadratic Equation Conditional Workflow

A routing graph that checks the discriminant of a quadratic equation and routes to different nodes: two real roots, one root, or complex roots.

Concepts: add_conditional_edges, router functions, branching logic in graphs

7️⃣ `7_review_response_conditional_workflow.ipynb` — Review Response Conditional Workflow

Takes a product review, classifies sentiment, then routes to a positive-response node or a negative-escalation node.

Concepts: LLM-powered routing, conditional branching, real-world classification use case

8️⃣ `8_x_post_generator_iterative_workflow.ipynb` — X Post Generator (Iterative Workflow)

Generates an X (Twitter) post, evaluates it, and loops back to refine if it doesn't meet quality criteria — demonstrating cycles in a graph.

Concepts: Cycles / loops in StateGraph, self-correcting agents, iteration counter, loop exit conditions

9️⃣ `9_basic_chatbot.ipynb` — Basic Chatbot

A simple conversational chatbot built with LangGraph's MessagesState, maintaining full message history across turns.

Concepts: MessagesState, multi-turn conversation, message accumulation, streaming

🔟 `10_persistence.ipynb` — Persistence & Checkpointing

Adds memory to the chatbot using a checkpointer so conversation state survives between sessions. Demonstrates thread-based state management.

Concepts: MemorySaver, SqliteSaver, thread_id config, state resumption, time-travel debugging

1️⃣4️⃣ `14_hitl.ipynb` — Human-in-the-Loop (HITL)

Pauses the graph mid-execution for human review or approval before continuing — the foundation of safe agentic systems.

Concepts: interrupt_before, interrupt_after, graph resumption, approval workflows, NodeInterrupt

1️⃣5️⃣ `15_subgraphs.ipynb` — Subgraphs

Composes complex systems by nesting graphs inside graphs — each subgraph has its own state and logic.

Concepts: Subgraph compilation, parent/child graph communication, modular agent design

1️⃣5️⃣ `15_subgraph_shared.ipynb` — Subgraphs with Shared State

Extends subgraph patterns by sharing state keys between parent and child graphs.

Concepts: Shared state schema, state inheritance, cross-graph data flow

⚡ `fault_tolerance.ipynb` — Fault Tolerance

Builds resilient workflows that handle node failures gracefully — retry logic, fallback nodes, and error state handling.

Concepts: Try/except in nodes, fallback edges, error state routing, retry patterns

🐍 Runnable Scripts

Script	Description
`chatbot_without_hitl.py`	Clean chatbot — continuous conversation, no interrupts
`chatbot_with_hitl.py`	Same chatbot with HITL interrupt — pauses for human approval before sensitive actions

Run either with:

python chatbot_without_hitl.py
python chatbot_with_hitl.py

⚙️ Setup & Installation

1. Clone the repo

git clone https://github.com/codeantik/langgraph.git
cd langgraph

2. Create a virtual environment

python -m venv venv
source venv/bin/activate        # macOS/Linux
venv\Scripts\activate           # Windows

3. Install dependencies

pip install langgraph langchain langchain-openai langchain-community python-dotenv openai jupyter

4. Set up environment variables

Create a .env file in the root:

OPENAI_API_KEY=your_openai_api_key_here
LANGCHAIN_API_KEY=your_langsmith_api_key_here   # optional, for tracing
LANGCHAIN_TRACING_V2=true
LANGCHAIN_PROJECT=langgraph-learning

5. Launch Jupyter

jupyter notebook

Then open notebooks in order, starting with 0_test_installation.ipynb.

🧠 Core Concepts Covered

Concept	Notebooks
`StateGraph` basics	`1`, `2`
Sequential chaining	`3`
Parallel execution (fan-out/in)	`4`
Conditional edges & routing	`6`, `7`
Iterative loops / cycles	`8`
Conversational chatbot	`9`
Persistence & checkpointing	`10`
Human-in-the-Loop (HITL)	`14`, `chatbot_with_hitl.py`
Subgraph composition	`15a`, `15b`
Fault tolerance & resilience	`fault_tolerance`

🗺️ Learning Roadmap

👨‍💻 Author

Ankit Singh — Full Stack Developer & Agentic AI Specialist

Also check out my LangSmith / LangChain fundamentals repo: codeantik/langsmith

📄 License

This project is open source and available under the MIT License.

Name		Name	Last commit message	Last commit date
Latest commit History 2 Commits
.gitignore		.gitignore
0_test_installation.ipynb		0_test_installation.ipynb
10_persistence.ipynb		10_persistence.ipynb
14_hitl.ipynb		14_hitl.ipynb
15_subgraph_shared.ipynb		15_subgraph_shared.ipynb
15_subgraphs.ipynb		15_subgraphs.ipynb
1_bmi_workflow.ipynb		1_bmi_workflow.ipynb
2_simple_llm_workflow.ipynb		2_simple_llm_workflow.ipynb
3_prompt_chaining_workflow.ipynb		3_prompt_chaining_workflow.ipynb
4_batsman_parallel_workflow.ipynb		4_batsman_parallel_workflow.ipynb
5_upsc_essay_workflow.ipynb		5_upsc_essay_workflow.ipynb
6_quad_eq_conditional_workflow.ipynb		6_quad_eq_conditional_workflow.ipynb
7_review_response_conditional_workflow.ipynb		7_review_response_conditional_workflow.ipynb
8_x_post_generator_iterative_workflow.ipynb		8_x_post_generator_iterative_workflow.ipynb
9_basic_chatbot.ipynb		9_basic_chatbot.ipynb
README.md		README.md
chatbot_with_hitl.py		chatbot_with_hitl.py
chatbot_without_hitl.py		chatbot_without_hitl.py
fault_tolerance.ipynb		fault_tolerance.ipynb

Folders and files

Latest commit

History

Repository files navigation

🕸️ LangGraph Learning Journey

📚 What This Repo Is

🗂️ File Structure

🧭 Notebook Breakdown

0️⃣ 0_test_installation.ipynb — Installation Check

1️⃣ 1_bmi_workflow.ipynb — BMI Workflow (Intro to StateGraph)

2️⃣ 2_simple_llm_workflow.ipynb — Simple LLM Workflow

3️⃣ 3_prompt_chaining_workflow.ipynb — Prompt Chaining Workflow

4️⃣ 4_batsman_parallel_workflow.ipynb — Batsman Parallel Workflow

5️⃣ 5_upsc_essay_workflow.ipynb — UPSC Essay Workflow

6️⃣ 6_quad_eq_conditional_workflow.ipynb — Quadratic Equation Conditional Workflow

7️⃣ 7_review_response_conditional_workflow.ipynb — Review Response Conditional Workflow

8️⃣ 8_x_post_generator_iterative_workflow.ipynb — X Post Generator (Iterative Workflow)

9️⃣ 9_basic_chatbot.ipynb — Basic Chatbot

🔟 10_persistence.ipynb — Persistence & Checkpointing

1️⃣4️⃣ 14_hitl.ipynb — Human-in-the-Loop (HITL)

1️⃣5️⃣ 15_subgraphs.ipynb — Subgraphs

1️⃣5️⃣ 15_subgraph_shared.ipynb — Subgraphs with Shared State

⚡ fault_tolerance.ipynb — Fault Tolerance

🐍 Runnable Scripts

⚙️ Setup & Installation

1. Clone the repo

2. Create a virtual environment

3. Install dependencies

4. Set up environment variables

5. Launch Jupyter

🧠 Core Concepts Covered

🗺️ Learning Roadmap

👨‍💻 Author

📄 License

About

Resources

Uh oh!

Stars

Watchers

Forks

Releases

Packages 0

Uh oh!

Contributors

Uh oh!

Languages

0️⃣ `0_test_installation.ipynb` — Installation Check

1️⃣ `1_bmi_workflow.ipynb` — BMI Workflow (Intro to StateGraph)

2️⃣ `2_simple_llm_workflow.ipynb` — Simple LLM Workflow

3️⃣ `3_prompt_chaining_workflow.ipynb` — Prompt Chaining Workflow

4️⃣ `4_batsman_parallel_workflow.ipynb` — Batsman Parallel Workflow

5️⃣ `5_upsc_essay_workflow.ipynb` — UPSC Essay Workflow

6️⃣ `6_quad_eq_conditional_workflow.ipynb` — Quadratic Equation Conditional Workflow

7️⃣ `7_review_response_conditional_workflow.ipynb` — Review Response Conditional Workflow

8️⃣ `8_x_post_generator_iterative_workflow.ipynb` — X Post Generator (Iterative Workflow)

9️⃣ `9_basic_chatbot.ipynb` — Basic Chatbot

🔟 `10_persistence.ipynb` — Persistence & Checkpointing

1️⃣4️⃣ `14_hitl.ipynb` — Human-in-the-Loop (HITL)

1️⃣5️⃣ `15_subgraphs.ipynb` — Subgraphs

1️⃣5️⃣ `15_subgraph_shared.ipynb` — Subgraphs with Shared State

⚡ `fault_tolerance.ipynb` — Fault Tolerance

Packages