RAG Chatbot Project

This project is a Retrieval-Augmented Generation (RAG) chatbot. It uses:

• Gemini 2.0 Flash as the language model (LLM) for generating answers
• Google BAAI embedding model for converting text to embeddings
• PostgreSQL with pgvector for storing and searching document embeddings

What is a RAG Chatbot?

A RAG chatbot answers user questions by first searching for relevant information in a set of documents, then sending both the user question and the found information to a language model (Gemini 2.0 Flash) to generate a helpful answer.

What You Need to Do

You will complete the code in each file to build a working RAG chatbot. Each file has TODO comments to guide you. Here is what each file/folder does and what you should implement:

Top-level files

• ingest.py: Script to read documents from the docs/ folder, generate embeddings using the embedding service, and store them in the vector database.
  • Implement: Document reading, embedding generation, and storing logic.

app/

This folder contains the main code, organized by responsibility:

• api/main.py: FastAPI app for the chatbot API.

  • Implement: Endpoints for /chat (get answer from chatbot) and /chat/history (get previous chats).

• chatbot/rag_pipeline.py: The core RAG pipeline logic.

  • Implement: Retrieve relevant context from the vector DB, call Gemini 2.0 Flash with the context and user query, and return the answer.

• core/config.py: Configuration settings.

  • Implement: Load API keys, DB connection info, and model names from environment variables.

• database/connection.py: Database connection code.

  • Implement: Connect to PostgreSQL using psycopg3, initialize the vector DB schema with pgvector.

• embeddings/embedding_service.py: Embedding service.

  • Implement: Use the Google BAAI model (via HuggingFace) to generate embeddings from text.

• retrieval/vector_store.py: Vector store logic.

  • Implement: Insert embeddings into Postgres/pgvector and retrieve the most similar documents for a query.

tests/

• test_api.py: Tests for the API endpoints.
  • Implement: Tests for root, /chat, and /chat/history endpoints.

docs/

• Add your source documents here. These will be ingested and used by the chatbot.

How to Run

1. Install dependencies: pip install -r requirements.txt
2. Add your documents to the docs/ folder.
3. Run the ingestion script: python ingest.py
4. Start the API: uvicorn app.api.main:app --reload
5. Use the API endpoints to chat with your RAG chatbot!

---

Tip: Each file contains TODO comments to help you know what to implement. Follow them step by step!