Hey Streamlit community!
I want to share ProData AI — an automated data science platform I built entirely with Streamlit.
What it does:
One-Click Mode — upload any CSV or Excel file and the entire pipeline runs automatically in 30 seconds:
→ Data cleaning
→ AutoML (6 models trained simultaneously)
→ 30-day Prophet forecasting
→ Business driver analysis (XAI)
→ Claude AI insights
→ PDF report generation
Manual Mode — full control over every step.
Tech stack:
- Streamlit for the entire UI
- scikit-learn for AutoML
- Prophet for forecasting
- Anthropic Claude API for AI chat
- Plotly for all charts
- fpdf2 for PDF generation
Some interesting Streamlit-specific decisions:
I used st.tabs() for the dashboard navigation instead of a custom sidebar which kept the layout clean and familiar to users.
The dashboard CSS uses st.markdown() with inline HTML for the KPI cards — this gave me full control over the white card style without needing custom components.
Live demo (no account, no signup):
Would love feedback from the community —
especially on performance with large datasets
and any Streamlit best practices I might be missing.
Check it out