← All courses
Agentic AI Live cohort Intermediate Flagship

Agentic AI Engineering with LangGraph & MCP

Design, build and ship multi-agent systems that do real work — safely.

SHIPSTACK · COURSE AGENTIC AI Agentic AIEngineering withLangGraph & MCP 8 MODULES · LIVE COHORT · INTERMEDIATE Flagship ~40h

About this course

From a single tool-calling agent to governed multi-agent workflows. You will learn the patterns that hold up in production — supervisors, handoffs, memory, human-in-the-loop and evaluation — and connect agents to real systems through the Model Context Protocol.

What you will be able to do

  • Build agents with tools, memory and structured outputs
  • Choose between supervisor, swarm and pipeline architectures
  • Expose and consume tools through MCP servers
  • Add guardrails, approvals and audit logs
  • Evaluate agents and deploy them as services

Curriculum

01Agents from first principles
  • The agent loop
  • Tool calling
  • Structured outputs
02LangGraph fundamentals
  • State and nodes
  • Conditional edges
  • Checkpointing
03Memory
  • Session vs. long-term memory
  • Semantic memory stores
  • Forgetting on purpose
04Multi-agent patterns
  • Supervisor
  • Handoffs and swarms
  • Debate and review
05Model Context Protocol
  • Writing an MCP server
  • Connecting clients
  • Security boundaries
06Human-in-the-loop & governance
  • Approvals and interrupts
  • Policy checks
  • Audit trails
07Evaluating agents
  • Trajectory evals
  • LLM-as-judge
  • Regression suites
08Shipping
  • Agents as APIs
  • Observability
  • Cost and latency control

Tools you will use

LangGraphLangChainMCPFastAPILangfuseDocker

Prerequisites

  • Comfortable with Python
  • Has called an LLM API before

Projects you will build

SHIPSTACK · PROJECT Wanderloom Multi-agent trip planner withLangGraph Intermediate Agentic AI 1 Traveller brief 2 Research agent 3 Planner agent 4 Itinerary 5 milestones
Agentic AI Intermediate

Multi-agent trip planner with LangGraph

Researcher, planner and budget agents collaborate to turn “5 days in Kerala in December under ₹40k” into a day-by-day itinerary.

LangGraph · Tavily · Streamlit Coming soon
SHIPSTACK · PROJECT ClaimMate Insurance claims copilot withlong-term memory Intermediate Agentic AI 1 Customer 2 Claims agent 3 Policy RAG 4 Claims API 5 milestones
Agentic AI Intermediate

Insurance claims copilot with long-term memory

Policyholders file and track claims in conversation; the agent remembers their policy, past claims and preferences across sessions.

LangGraph · LangMem · pgvector Coming soon
SHIPSTACK · PROJECT RefundRadar Governed multi-agent returns &refund triage for e-commerce Advanced Agentic AI 1 Return request 2 Policy agent 3 Risk agent 4 Reviewer desk 5 milestones
Agentic AI Advanced

Governed multi-agent returns & refund triage for e-commerce

A team of agents reviews every return request, auto-approves the clean ones and routes risky or expensive cases to a human fraud desk.

LangGraph · FastAPI · PostgreSQL Coming soon
SHIPSTACK · PROJECT TicketZero IT support copilot withagentic RAG Advanced Agentic AI 1 Employee 2 Router agent 3 Web search 4 Answer or ticket 5 milestones
Agentic AI Advanced

IT support copilot with agentic RAG

Resolve “my VPN won’t connect” before it becomes a ticket: an agent that searches the internal knowledge base first and the web only when it has to.

LangGraph · FastAPI · Pinecone Coming soon