Autonomous Finance Tutor

Production-structured, beginner-friendly project using:

• Gemini API (Google AI Studio)
• CrewAI (Planner + Tutor agents)
• LangChain + ChromaDB (RAG)
• Streamlit frontend
• Docker
• GitHub Actions CI
• Kubernetes deployment

Project Structure

src/

• agents/
  • planner.py
  • tutor.py
• tools/
  • retriever.py
• utils/
  • logger.py
• app.py

Local Run

1. Install dependencies:

   pip install -r requirements.txt

2. Start Streamlit app:

   streamlit run src/app.py

3. Open app in browser (usually http://localhost:8501).

4. In sidebar:

   • Add Gemini API key
   • Upload finance PDF
   • Click Ingest PDF

5. Ask a finance question in chat.

Docker Run

1. Build image:

   docker build -t fin-agent .

2. Run container:

   docker run -p 8501:8501 fin-agent

Kubernetes Run (Minikube or K3d)

1. Build image locally for your cluster runtime.

2. Apply manifest:

   kubectl apply -f k8s/deployment.yaml

3. Check resources:

   kubectl get pods kubectl get svc

CI

GitHub Actions workflow at .github/workflows/ci.yml runs:

• Ruff lint on src
• Docker build verification

Error Handling Implemented

• Missing API key -> Streamlit warning
• No PDF uploaded -> Streamlit alert
• Empty retrieval -> fallback warning and general tutor response
• API failures -> one retry with error shown on failure