Tangle: A Visual Editor for ML Pipelines

Tangle is an open-source, platform-independent environment designed for the collaborative creation of machine learning and data pipelines. It features a robust drag-and-drop interface that eliminates the need for complex initial configurations the requirement for tedious setup.

Our Mission: Developed by the software and ML engineering teams at Shopify, Tangle aims to democratize the way machine learning pipelines are constructed and managed.

🚀 Core Capabilities for ML Teams

Tangle provides a suite of tools to streamline the experimentation process:

1. Visual Pipeline Orchestration

Users can architect intricate ML workflows using a powerful visual canvas.

• Drag & Drop: Easily arrange components.
• Configuration: Define arguments and link components together.
• Flexibility: Build your entire flow with or without writing manual code.

2. Collaborative Development

Tangle is built for teamwork. You can clone, modify, and execute any existing pipeline to suit your specific needs.

• Component Library: Utilize a set of pre-made, reusable modules.
• Custom Extensions: Integrate your own logic by wrapping existing code into a container using the integrated in-app editor.

3. Optimized Execution

To reduce waste and increase speed, Tangle implements Advanced Execution Caching. By utilizing content-based caching, the system avoids redundant computations.

The Efficiency Logic: If the input content hasn't changed, the system retrieves the cached result: $\text{Execution Time} = \sum (\text{Step}_n) - \text{Cached Steps}$

---

🛠 Pipeline Workflow Visualization

The following diagram illustrates how a typical Tangle workflow operates:

🏁 Getting Started

Ready to dive in? Follow these steps to begin your journey:

• Explore the Example Pipelines to see Tangle in action.
• Review the official Documentation.
• Reach out to the team for support or guidance.

---

💻 Open Source & Contribution

Tangle is fully open source. We encourage the community to provide feedback, report bugs, and contribute to the codebase.

Repository Access

Example Configuration

If you are adding a custom component, your configuration might look like this:

{
  "component": "TangleCustomModule",
  "params": {
    "learning_rate": 0.001,
    "batch_size": 32,
    "cache_enabled": true
  }
}

Have questions? Our team is standing by to help you master visual ML pipeline development!