🕸️Gen AI
LangGraph
Stateful multi-agent graph workflows
Build complex stateful AI agent systems using LangGraph. Covers graph-based workflows, multi-agent orchestration, checkpointing, human-in-the-loop, and production deployment.
36 lessons
08 hrs
Advanced
What you will learn
Build real-world projects from scratch
Write clean, production-ready code
Understand core concepts deeply with diagrams
Follow industry best practices
Get hands-on with code in every lesson
Access lifetime updates as the tech evolves
Curriculum
Module 1What LangGraph Is, and Why It Exists Alongside LangChain2 lessons
Module 2State Schema and Reducers2 lessons
Defining State With TypedDict — And Why a Plain Field Gets Overwritten12 min
Reducers — Telling LangGraph How to Combine Updates Correctly12 min
Module 3Nodes, Edges, and .compile() — Building a Real, Running Graph2 lessons
Nodes Are Just Functions That Read and Return State12 min
Wiring Nodes Into a Real, Compiled Graph13 min
Module 4Conditional Edges and Cycles — The Real ReAct Loop From Scratch2 lessons
add_conditional_edges — Routing Based on Real State, Not a Fixed Sequence13 min
Building a Real Cycle — The ReAct Loop From Scratch14 min
Module 5Checkpointers and thread_id — Real Persistence Across Separate Runs2 lessons
Confirming Module 4's Graph Is Still Stateless Between Separate Calls11 min
Attaching a Real Checkpointer — Persistence Across Separate Calls13 min
Module 6Time-Travel Debugging — Inspecting and Replaying Real State History2 lessons
get_state_history — Every Single Step Was Already Saved12 min
Rewinding and Resuming From an Earlier Checkpoint13 min
Module 7Production Persistence — MemorySaver vs SqliteSaver vs PostgresSaver2 lessons
Measuring InMemorySaver's Real Limitation — Data Lost on Restart12 min
SqliteSaver vs PostgresSaver — An Honest, Evidence-Based Choice12 min
Module 8interrupt() and Human Approval Gates2 lessons
Why a Genuine Pause Requires Real Persistence, Not Just a Python Function Call12 min
Building a Real Approval Gate With interrupt()14 min
Module 9Streaming Tokens, State Updates, and Intermediate Steps2 lessons
stream_mode='updates' — Watching Each Node's Contribution As It Happens12 min
Streaming Model Tokens and the Full Values Mode13 min
Module 10create_react_agent — The Fast, High-Level Path2 lessons
Confirming create_react_agent Produces the Same Structure You Already Built12 min
What create_react_agent Lets You Customize, and What It Hides12 min
Module 11ToolNode and Parallel Tool Execution2 lessons
ToolNode — A Real, Directly Importable Building Block12 min
Parallel Tool Execution — Measuring ToolNode's Real Concurrency13 min
Module 12When to Graduate From Prebuilt to Manual StateGraph2 lessons
Constructing a Task create_react_agent Genuinely Cannot Express13 min
Building the Fix With a Hand-Built StateGraph13 min
Module 13The Supervisor Pattern — One Agent Routing to Specialists2 lessons
The Supervisor Pattern, Defined12 min
A Three-Specialist Supervisor With a Genuine Fallback Path13 min
Module 14Parallel Fan-Out and Result Aggregation2 lessons
Fan-Out and Fan-In, Defined12 min
Measuring Fan-Out's Real Concurrency, With Real Timing12 min
Module 15Multi-Agent Handoffs and Shared State Design2 lessons
A Handoff, Defined13 min
A Real, Bidirectional Handoff and Designing Shared State14 min
Module 16Node-Level Error Handling, Timeouts, and Recovery2 lessons
RetryPolicy, Defined — And What Happens Without One12 min
Graceful Degradation When Retries Are Exhausted13 min
Module 17Deploying a LangGraph Application2 lessons
Load-Once Serving With Real, Checkpointer-Backed Persistence13 min
Streaming the Served Graph Through a Real Endpoint12 min
Module 18LangChain vs LangGraph, Revisited Honestly2 lessons
The Three-Question Framework, Revisited With Real Evidence12 min
Using Both Tools Together in One Real Graph13 min
36 total lessons across 18 modulesPreview Module 1 free