Agentic Systems Practice

This repository contains a curated collection of Jupyter notebooks and Python scripts created while learning to build agentic systems using LangChain and LangGraph.
The goal is to explore and implement core components of LLM-based agents such as:

• Memory
• Retrieval-Augmented Generation (RAG)
• Tool usage (prebuilt & custom)
• Agent orchestration with the new LangChain Expression Language (LCEL)
• LangGraph-powered workflows for more control and state management

---

📂 Repository Contents

1. Agentic Systems with LangChain

Script	Description
script1_aa_ai_assistant.ipynb	A simple AI assistant built with LangChain, demonstrated with and without memory integration.
script2_aa_rag_lcel.ipynb	Implements Retrieval-Augmented Generation (RAG) using LCEL for building modular chains.
script3_aa_tools_agent_tool_calling.ipynb	Introduces tool usage in LangChain v0.1, including prebuilt and custom tools, and building a tool-calling agent.
script4_aa_tools_react.ipynb	Demonstrates the ReAct (Reasoning + Acting) framework with prebuilt and custom tools in LangChain v0.1.
script5_aa_tools_agent_self_ask_with_search.ipynb	Implements the Self-Ask-With-Search agent pattern for step-by-step reasoning with a search tool.

---

2. Agentic Systems From Scratch (Pure Python)

Script	Description
script6_aa_agent_react_from_scratch.ipynb	Implements the ReAct agent from scratch to deeply understand each functionality.

---

3. Agentic Systems with LangGraph

Script	Description
script7_aa_agent_langgraph_introduction.ipynb	Introduction to LangGraph with key concepts and terminology.
script8_aa_agent_langgraph_agent_with_RAG.ipynb	Building a ReAct agent with RAG using LangGraph.
script9_aa_agent_langgraph_checkpoint_memory.ipynb	Demonstrates how to define memory checkpoints for LangGraph agents.

---

4. chatbot_with_langgraph/ — Complete Web-Based Chatbot Agent

Script	Description
aa_app.py	Streamlit-based chatbot interface for interacting with the LangGraph agent.
aa_bot.py	Core chatbot logic — routes input through: input → agent → tool → end or input → agent → tool → agent.

---

📝 Notes

• More scripts coming soon.
• All scripts are annotated and self-contained for ease of understanding and modification.
• Suitable for:
  • Beginners exploring LLM-based agent design
  • Developers experimenting with LangChain and LangGraph
  • Researchers prototyping tool-using AI agents
• Inspired by tutorials from the LangGraph Agents YouTube Playlist.

---