RAG Chatbot with LangChain
Based on my Master's degree research, I built a chatbot using a Retrieval-Augmented Generation (RAG) pipeline with LangChain to allow anyone to interact directly with my thesis findings through simple questions.
I converted my thesis from LaTeX to Markdown, segmented it into smaller chunks, stored these locally in a Chroma vector database and embedded each segment using the E5-Large model. When a user submits a query, LangChain retrieves the most relevant chunks from the database and passes them to 4o-mini along with a structured prompt template, returning context-rich responses.
The chatbot’s interface is built with Streamlit and runs on a DigitalOcean droplet using Caddy as a reverse proxy. If you’re interested in trying the chatbot and interacting directly with my thesis, feel free to reach out to me to obtain an access code.