RAG-powered CV Chatbot

The idea is simple: you upload any CV (PDF) and start asking questions about it, and the AI answers based only on the information inside the CV.

I built it while learning LangChain and RAG, and it was a great way to understand how these systems actually work in practice.

Tech stack I used:

LangChain to connect the whole pipeline

Groq + LLaMA 3.1 (8B) as the LLM for generating responses

HuggingFace Embeddings (paraphrase-multilingual-MiniLM-L12-v2) to convert text into vectors

FAISS as the vector database

PyPDF to extract text from the CV

RecursiveCharacterTextSplitter to split the CV into chunks (500 chars with 50 overlap)

Conversation Memory so the chatbot remembers the context of the conversation

How it works:

Upload a CV as a PDF

Extract the text from the file

Split the text into smaller chunks

Convert the chunks into embeddings

Store them in FAISS

When a question is asked, the retriever finds the most relevant chunks

The LLM uses those chunks + the conversation history to generate the answer

Everything was built on Google Colab using open-source tools.