Custom Data Chatbot

Custom Data Chatbot is a Streamlit-based RAG application for chatting with uploaded PDF course materials. It supports:

• PDF upload and indexing
• semantic chunking and FAISS retrieval
• document-grounded answers
• general-knowledge fallback when retrieval is weak
• short-term chat memory
• separate Chat, Workspace, and Settings pages

The repository is structured to run locally and deploy on Streamlit Community Cloud.

##Streamlit App Link ------- https://furgan0101-chatbot-appstreamlit-app-wj0wva.streamlit.app/

Project Structure

custom-data-chatbot/
├── app/
│   ├── pages/
│   │   ├── 1_Workspace.py
│   │   └── 2_Settings.py
│   ├── streamlit_app.py
│   └── ui_shared.py
├── data/
│   └── docs/
├── src/
│   ├── chatbot.py
│   ├── chunking.py
│   ├── embed.py
│   ├── ingest.py
│   ├── rerank.py
│   ├── retrieve.py
│   ├── settings.py
│   └── utils.py
├── tests/
├── .streamlit/
│   └── secrets.toml
├── requirements.txt
├── .gitignore
└── README.md

Run Locally

1. Create and activate a Python 3.11+ virtual environment.
2. Install dependencies:

pip install -r requirements.txt

3. Add your OpenAI API key to .streamlit/secrets.toml:

OPENAI_API_KEY = "your_api_key_here"

4. Start the app:

streamlit run app/streamlit_app.py

Local Configuration

Non-secret runtime defaults can be adjusted in .env.example and copied into your shell environment if needed. The OpenAI API key should not be stored in .env; the app reads it from st.secrets["OPENAI_API_KEY"].

The Settings page lets you change provider, model, and RAG behavior during the current session.

Streamlit Community Cloud Deployment

1. Push this repository to GitHub.
2. In Streamlit Community Cloud, create a new app pointing to:
   • repository: this repo
   • main file path: app/streamlit_app.py
3. In the Streamlit Cloud app settings, add:

OPENAI_API_KEY = "your_api_key_here"

4. Deploy.

Deployment Notes

• The recommended cloud provider mode is OpenAI.
• Ollama requires a local Ollama service and is not available on Streamlit Community Cloud.
• Uploaded PDFs are stored inside the app workspace at runtime. Re-deploys or container resets may clear runtime-generated files unless they are re-uploaded.
• sentence-transformers downloads model weights on first use, so the first indexing run may be slower.

Testing

python -m unittest tests.test_hybrid_chatbot

Secrets

The app reads the OpenAI key through Streamlit secrets:

st.secrets["OPENAI_API_KEY"]

Demo