LangSmith Masterclass is a comprehensive, hands-on repository designed to teach and demonstrate advanced concepts in building LLM-powered applications using LangChain, LangSmith, and LangGraph. The project covers:

• Simple LLM prompt chaining
• Sequential multi-step chains
• Retrieval-Augmented Generation (RAG) pipelines with PDF data
• Custom agents with web search and API tools
• Graph-based workflows for structured evaluation

Each script is crafted for clarity and experimentation, making this repo ideal for learners, researchers, and developers who want to master modern AI application development.

• Project Structure
• Setup Instructions
• Environment Variables
• Detailed File Explanations
• Requirements
• Usage Guide
• Troubleshooting
• References

langsmith-masterclass/
├── 1_simple_llm_call.py
├── 2_sequential_chain.py
├── 3_rag_v1.py
├── 3_rag_v2.py
├── 3_rag_v3.py
├── 3_rag_v4.py
├── 4_agent.py
├── 5_langgraph.py
├── islr.pdf
├── requirements.txt
├── myvenv/
│   └── ... (virtual environment files)
└── .gitignore

• 1_simple_llm_call.py: Basic LLM prompt chaining example.
• 2_sequential_chain.py: Sequential multi-step LLM chain for report generation and summarization.
• 3_rag_v1.py - 3_rag_v4.py: Progressive Retrieval-Augmented Generation (RAG) pipelines using PDFs, embeddings, and FAISS vector stores. Each version adds more features (tracing, caching, modularity).
• 4_agent.py: Custom agent using LangChain's ReAct framework, with web search and weather data tools.
• 5_langgraph.py: Advanced graph-based workflow using LangGraph for structured evaluation and feedback.
• islr.pdf: Sample PDF for RAG demos.
• requirements.txt: All required Python packages.
• myvenv/: Local Python virtual environment.
• .gitignore: Standard ignores for Python, virtualenv, Jupyter, and VS Code files.

1. Clone the repository

   git clone <repo-url>
   cd langsmith-masterclass

2. Create and activate a virtual environment

   python -m venv myvenv
   .\myvenv\Scripts\Activate.ps1

3. Install dependencies

   pip install -r requirements.txt

4. Set up environment variables

   • Create a .env file in the root directory with your API keys:

     OPENAI_API_KEY=your_openai_key
     LANGCHAIN_API_KEY=your_langchain_key
     LANGCHAIN_TRACING_V2=true
     LANGCHAIN_PROJECT=your_project_name

• OPENAI_API_KEY: Required for OpenAI LLM and embedding models.
• LANGCHAIN_API_KEY: Required for LangSmith tracing and analytics.
• LANGCHAIN_TRACING_V2: Enables advanced tracing (recommended: true).
• LANGCHAIN_PROJECT: Project name for LangSmith dashboard.

• Loads environment variables.
• Sets up a simple prompt and runs a question through an OpenAI chat model.

• Demonstrates a multi-step chain: generates a report, then summarizes it.
• Uses two different LLM models and prompt templates.
• Shows how to pass config metadata for tracing.

• Loads a PDF, splits it into chunks, embeds with OpenAI, and builds a FAISS vector store.
• Runs a RAG pipeline to answer questions from the PDF context.

• Adds LangSmith tracing to each pipeline step using @traceable decorators.
• Shows how to organize and trace modular pipeline functions.

• Further modularizes the RAG pipeline and traces parent setup functions.
• Demonstrates advanced tracing and pipeline composition.

• Adds caching, fingerprinting, and index management for efficient RAG workflows.
• Uses Path, hashlib, and JSON for robust file handling.

• Implements a ReAct agent with custom tools (web search, weather API).
• Uses LangChain Hub for prompt templates.
• Demonstrates agent creation, tool registration, and invocation.

• Builds a graph-based workflow for essay evaluation using LangGraph.
• Defines structured schemas and feedback models.
• Shows how to use state graphs and custom evaluation logic.

All dependencies are listed in requirements.txt. Key packages include:

• langchain, langchain-openai, langchain-community, langsmith, langgraph
• faiss-cpu, pypdf, python-dotenv, openai, pandas, streamlit, etc.
• See the full list in requirements.txt for exact versions.

• Run any script directly after activating the virtual environment:

  python 1_simple_llm_call.py
  python 2_sequential_chain.py
  python 3_rag_v1.py
  # ... and so on

• For RAG scripts, ensure islr.pdf is present in the root directory.
• For agent scripts, internet access is required for API calls.
• For LangSmith tracing, ensure your API keys and project name are set in .env.

• Missing packages: Run pip install -r requirements.txt again.
• API errors: Check your .env file for correct keys.
• PDF not found: Ensure islr.pdf exists in the root directory.
• Virtual environment issues: Recreate with python -m venv myvenv.
• Windows activation: Use Activate.ps1 for PowerShell.

• LangChain Documentation
• LangSmith Documentation
• LangGraph Documentation
• OpenAI API

For questions or contributions, please open an issue or pull request.