A full-day, hands-on repository for learning to design, build, and ship Agentic AI systems using LangGraph.
The modules progress from LLM I/O and prompting to tool-use, routing, memory and context engineering, advanced multi-agent systems, Agentic RAG and Multimodal Agentic AI, and finally deployment, monitoring, and evaluation.
Workshop instructor: Dipanjan (DJ) Sarkar
Module 1: Introduction to Generative AI and Agentic AI
Module 2: Building Simple Agentic AI Systems
Module 3: Context Engineering for Agentic AI Systems
Module 4: Building Advanced Agentic AI Systems
Module 5: Building Agentic RAG and Multimodal Agentic AI Systems
Module 6: Deploying, Monitoring, and Evaluating Agentic AI Systems
Get everyone on the same page: a fast recap of GenAI concepts, standard LLM input/output patterns, and prompting basics.
Then move into LangGraph fundamentals—state, nodes, edges and add simple control with conditional routing.
You’ll finish by building a minimal chatbot, a standard RAG pipeline, and hands-on tool calling to prepare for later agent systems.
-
M1LC1 - LLM Input/Output and Prompting
·
-
M1LC2 - Simple Agentic Graph and State Management in LangGraph
-
M1LC3 - Conditional Routing in LangGraph
-
M1LC4 - Build a LLM-powered Chatbot with LangGraph
-
M1LC5 - Build a RAG System with LangGraph
-
M1LC6 - Implementing Tools & Tool Calling for Agentic AI Systems
Go from prompting to practical agents that call tools using the ReAct pattern. You will build a LangGraph ReAct agent, recreate the same flow in CrewAI to compare design tradeoffs, and then implement a Text2SQL assistant that plans, generates SQL, runs queries on a sample database, and explains results with guardrails.
-
M2LC1 - Build a Tool Use ReAct Agentic AI System with LangGraph
-
M2LC2 - Build a Tool Use ReAct Agentic AI System with CrewAI
-
M2LC3 - Build a Text2SQL Agentic AI System with LangGraph
Design memory-aware agents that work across users and sessions. Learn short-term versus long-term memory, threads and snapshots, and context window management. Use LangMem to persist and retrieve context. Add MCP servers to expose external tools and data. Build a multi-user conversational financial analyst and an adaptive agent that learns from past sessions.
-
M3LC1 - Building Agentic AI Systems with Pre built MCP Servers
-
M3LC2 - Building Agentic AI Systems with MCP Servers and Clients from Scratch
-
M3LC3 - Build a Multi User Conversational Agentic AI Financial Analyst with Memory Context Engineering
-
M3LC4 - Build a Multi User Multi Session Adaptive Agentic AI Systems with Short and Long Term Memory
Go beyond basic tool-use and add planning, reflection, and collaboration across agents. Learn when to choose single agent versus multi agent patterns, how to parallelize plan execution for speed, and how to structure supervisor and specialist roles. Build a deep research agent that uses tools and the web, and a multi agent workflow for utilization review.
-
M4LC1 - Parallelized Plan Execution in Report Planner Agents
-
M4LC2 - Build a Deep Research Agentic AI System
-
M4LC3 - Build a Multi Agent System for Utilization Review
Combine retrieval with agents to route questions to the right data and tools. Learn Router Agentic RAG to build a customer support workflow that selects collections, tools, and prompts based on query intent. Jump to multimodal agentic ai by processing invoices with multimodal LLMs and structured extraction, coordinating multiple agents for parsing, validation, and final summaries.
-
M5LC1 - Build a Customer Support Router Agentic RAG System
-
M5LC2 - Build a Multimodal Multi Agent System for Invoice Processing
Turn a working agent into a reliable web service API. Package a LangGraph agent behind a FastAPI endpoint, run it on the cloud, and support for API calling and async live streaming. Instrument the system with tracing, monitoring and observability metrics, and evaluation runs. Use LangSmith to monitor latency, cost, tool calls, errors, and agent goal, reasoning and trajectory quality.
-
M6LC1 - Build and Deploy a Healthcare Utilization Review AI Agent with LangGraph and FastAPI
-
M6LC2 - Monitoring and Evaluating Agentic AI Systems
This workshop builds on open source work and docs: