Code Wizard is a super cool AI chatbot that helps you learn and use the LangChain Documentation in an interactive way. Just ask it anything about LangChain concepts or code, and it'll break it down for you in an easy-to-understand way. Built with Next.js, FastAPI, LangChain, LangGraph, LCEL, and ChatOpenAI which can be easily switched a local LLaMA model.
Link to project: https://code-wizard-frontend.vercel.app/
Frontend Repo: https://github.com/RutamBhagat/code_wizard_frontend
Backend Repo: https://github.com/RutamBhagat/Code-Wizard-LangGraph-C_RAG
Code.Wizard.Walkthrough.ELlHz7vGVGQ.mp4
Tech used: LangChain, LangGraph, LCEL, LLaMA, Next.js, Typescript, FastAPI
Code Wizard is a full-stack app that combines some cutting edge tech. The front-end is built with Next.js and React, which makes it fast and snappy. The back-end uses FastAPI to host the LangChain and LangGraph pipelines and APIs.
The core is powered by LangChain, which lets us build cool applications with large language models like ChatOpenAI. Code Wizard takes the LangChain documentation, chunks it up, and stores it in a vector database using embeddings.
When you ask Code Wizard a question, it uses a LangGraph DAG to search the vector database for relevant doc chunks. It then generates a response by combining those chunks with the ChatOpenAI.
The front-end has a slick chat interface built with React, so you can have natural conversations with the AI. It also renders Markdown and code snippets nicely.
To make Code Wizard performant and scalable, it uses some cool optimizations:
- Caching: Frequently asked questions and their responses are cached to improve latency.
- Streaming Response: The LangChain pipelines to stream the response to the user.
- Model Optimization: Switching the model to a better but more expensive model like gpt-4o.
Building Code Wizard was an incredible learning experience that taught me so much about LangChain, LangGraph, LCEL, large language models, and modern web development. Some key lessons:
- How to effectively integrate LangChain components like agents, memory, chat history, agent state, nodes, chains, and vector stores.
- Optimizing LLM performance through techniques like quantization and CPU offloading.
- Designing intuitive conversational UIs that feel natural and engaging.
- Leveraging the latest web frameworks like Next.js and FastAPI for building scalable apps.
Check out these example conversations with Code Wizard:
What is Langchain:
Explaining LangChain Expression Language:
