business-science/ai-data-science-team

AI Data Science Team

AI Data Science Team is a Python library of specialized agents for common data science workflows, plus a flagship app: AI Pipeline Studio. The Studio turns your work into a visual, reproducible pipeline, while the AI team handles data loading, cleaning, visualization, and modeling.

Status: Beta. Breaking changes may occur until 0.1.0.

Please ⭐ us on GitHub (it takes 2 seconds and means a lot).

AI Pipeline Studio (Flagship App)

AI Pipeline Studio is the main example of the AI Data Science Team in action.

Highlights:

• Pipeline-first workspace: Visual Editor, Table, Chart, EDA, Code, Model, Predictions, MLflow
• Manual + AI steps with lineage and reproducible scripts
• Multi-dataset handling and merge workflows
• Project saves: metadata-only or full-data
• Storage footprint controls and rehydrate workflows

Run it:

streamlit run apps/ai-pipeline-studio-app/app.py

Full app docs: apps/ai-pipeline-studio-app/README.md

Quickstart

Requirements

• Python 3.10+
• OpenAI API key (or Ollama for local models)

Install the app and library

Clone the repo and install in editable mode:

pip install -e .

Run the AI Pipeline Studio app

streamlit run apps/ai-pipeline-studio-app/app.py

Library Overview

The repository includes both the AI Pipeline Studio app and the underlying AI Data Science Team library. The library provides agent building blocks and multi-agent workflows for:

• Data loading and inspection
• Cleaning, wrangling, and feature engineering
• Visualization and EDA
• Modeling and evaluation (H2O + MLflow tools)
• SQL database interaction

Agents (Snapshot)

Agent examples live in examples/. Notable agents:

• Data Loader Tools Agent
• Data Wrangling Agent
• Data Cleaning Agent
• Data Visualization Agent
• EDA Tools Agent
• Feature Engineering Agent
• SQL Database Agent
• H2O ML Agent
• MLflow Tools Agent
• Multi-agent workflows (e.g., Pandas Data Analyst, SQL Data Analyst)
• Supervisor Agent (oversees other agents)
• Custom tools for data science tasks

Apps

See all apps in apps/. Notable apps:

• AI Pipeline Studio: apps/ai-pipeline-studio-app/
• EDA Explorer App: apps/exploratory-copilot-app/
• Pandas Data Analyst App: apps/pandas-data-analyst-app/

Use OpenAI

from langchain_openai import ChatOpenAI
llm = ChatOpenAI(
    model_name="gpt-4.1-mini",
)

Use Ollama (Local LLM)

ollama serve
ollama pull llama3.1:8b

from langchain_ollama import ChatOllama

llm = ChatOllama(
    model="llama3.1:8b",
)

Next-Gen AI Agentic Workshop

Want to learn how to build AI agents and AI apps for real data science workflows? Join my next‑gen AI workshop: https://learn.business-science.io/ai-register