Agentic & No-Code Data Transformations
Vibe coded pipelines: say hello to accuracy and maintainability.

Website · Documentation · Issues · Contributing

What is Visitran?

Visitran is the modern, agent-native data transformation platform that lets you create sophisticated, yet highly standardized transformation metrics and highly maintainable transformation projects with AI. It is built on a strong foundation of Python (powered by Ibis) and SQL, while affording you a simple, conversational interface.

I just vibe code my pipelines. Why do I need Visitran?

First off, congratulations. You are using modern tools to quickly solve pains you have. The trouble starts when everyone in your organization is vibe coding pipelines and seeing different numbers for critical metrics. Pandemonium ensues. As it is, it's tough for even humans to come to an agreement on how core metrics are defined. With Visitran, you get the best of both worlds. You "vibe code" your pipelines with Visitran AI, but you can collaborate with your team members and build transformations that are agreed upon, verifiable and maintainable. Speed: say hi to accuracy. Accuracy: say hi to maintainability.

Supported Databases

If Visitran helps you, please star this repo — it helps others discover the project!

Key Features

For Data Engineers

• No-code backed by Python-based transformation models with Ibis SQL generation
• 6 database adapters with table, view, and incremental materialization
• DAG execution engine with dependency resolution
• Job scheduling with cron/interval triggers, retry, and chaining

For Analysts & No-Code Users

• No-code model builder — joins, filters, aggregates, window functions, pivots, unions
• Formula Columns support: Create complex transformations with Excel-like formulas
• AI assistant for natural language to data transformations
• Built-in testing framework for model validation

Getting Started

Requires Docker and Docker Compose.

git clone https://github.com/Zipstack/visitran.git
cd visitran
cp backend/sample.env backend/.env
cd docker
docker compose up --build -d

Open http://localhost:3000 and Sign Up to create your account — that's it!

AI Features (optional): Get an API key from app.visitran.com and add VISITRAN_AI_KEY=vtk_... to backend/.env, then restart with docker compose restart backend.

git clone https://github.com/Zipstack/visitran.git
cd visitran
cp backend/sample.env backend/.env

cd backend
pip install uv
uv sync
source .venv/bin/activate
python manage.py migrate
python manage.py runserver

cd frontend
cp sample.env .env
npm install
npm start

Project Structure

backend/
├── backend/                     # Django application
│   ├── core/                    #   Models, views, REST API
│   ├── application/             #   Business logic, config parser, interpreter
│   └── server/settings/         #   Django settings
├── visitran/                    # Core transformation engine (standalone)
│   ├── adapters/                #   Database adapters (postgres, duckdb, snowflake, etc.)
│   ├── templates/               #   VisitranModel base class
│   └── visitran.py              #   DAG execution engine

frontend/src/
├── base/                        # App shell, routing, project listing
├── ide/                         # Main workspace (editor, file explorer, SQL drawer)
│   ├── chat-ai/                 #   AI assistant interface
│   └── editor/                  #   No-code visual model builder
├── store/                       # Zustand state management
├── service/                     # API services (axios, socket)
└── widgets/                     # Reusable UI components

docker/
├── dockerfiles/                 # Backend and frontend Dockerfiles
└── docker-compose.yaml          # Docker Compose configuration

Development

Backend

# Run server with hot reload
cd backend && python manage.py runserver

# Run migrations
python manage.py migrate

# Lint & type check
inv checks
inv type_check

# Run tests (requires Docker for test databases)
docker compose up --wait
uv run pytest -vv --dist loadgroup -n 5 tests
docker compose down

Frontend

cd frontend

npm start              # Dev server
npm run build          # Production build
npm test               # Run tests
npm run test:coverage  # Test with coverage
npm run lint           # ESLint check
npm run lint:fix       # ESLint auto-fix

Cloud & Enterprise

For teams that need managed infrastructure, advanced features, or enterprise-grade security.

• SSO & RBAC — SAML/OIDC single sign-on with role-based access control
• Managed AI — Built-in AI service with no API key setup required
• Job Scheduling — Cron/interval triggers with retry, chaining, and notifications
• Cloud Storage — Managed file storage with version control
• Multi-Tenant — Organization-level isolation and team management
• Priority Support — Dedicated support with SLA guarantees

The AI assistant generates data transformations from natural language. In OSS mode, AI access requires a Visitran AI subscription key:

1. Sign up at app.visitran.com
2. Subscribe to an AI plan and generate an API key
3. Go to Settings > Visitran AI in your local instance and enter your key
4. Open the Chat AI drawer in the IDE to start using AI

The cloud gateway validates your key, routes prompts through LLM models (Claude, GPT-4o, Gemini), and streams responses back — no local AI setup or LLM credentials needed.

Tech Stack

Contributing

We welcome contributions! See CONTRIBUTING.md for detailed guidelines.

1. Fork the repo and create a feature branch from main
2. Make your changes
3. Run linting:
   • Frontend: npm run lint
   • Backend: inv checks
4. Run tests:
   • Frontend: npm test
   • Backend: uv run pytest
5. Submit a pull request

Community

Join the Visitran community:

A Note on Analytics

Visitran integrates Posthog to track minimal usage analytics. Disable by setting REACT_APP_ENABLE_POSTHOG=false in the frontend's .env file.

License

Visitran is licensed under the GNU Affero General Public License v3.0 (AGPL-3.0). You are free to use, modify, and distribute this software under the terms of the AGPL-3.0. If you run a modified version as a network service, you must make the source code available to users of that service.