OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
AI Developer Tools & SDKs

Tangle

Tangle is an open-source Python framework and web application for visually building, orchestrating and running machine-learning experiments and production pipelines. It runs containerized command-line tasks as connected workflows and supports local Docker or Podman use as well as execution through Hugging Face Jobs.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Tangle — OpenViST project profile in AI Developer Tools & SDKs.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use Tangle when you need a visual way to assemble and run repeatable machine-learning or data-processing workflows, and want a choice between local and Hugging Face execution.

Who it is for

  • ML engineers who want to build and iterate on experiment or production workflows.
  • Teams that need to assemble and monitor reusable pipeline tasks.
  • Users who want to design pipelines visually, including those who do not want to define the workflow entirely in code.

Documented uses

  • Connect pipeline components and configure task arguments in a visual editor.
  • Orchestrate batch workflows such as training or data processing.
  • Submit runs and monitor task progress, logs and artifacts.
  • Clone prior runs for further editing and reuse cached task executions where possible.
Before you choose

Requirements & limits

The Hugging Face interface allows pipeline creation without registration. Submitting a run requires login and a Hugging Face Pro subscription. The README lists Docker and uv for local setup.

  • The shared Hugging Face deployment executes each user’s pipelines through that user’s Hugging Face Jobs and stores logs and artifacts in the user’s private Hugging Face Dataset repository.
  • A single-tenant Hugging Face Space can be duplicated to an account; an organization deployment supports multiple team members viewing one another’s pipeline runs.
  • The visual workflow model orchestrates containerized command-line programs rather than functions passing shared in-memory objects.
  • The project uses the ComponentSpec format and supports reuse of Kubeflow and Vertex AI pipeline components.
  • Run graphs, logs and artifact metadata are retained, and content-based caching can reuse matching task executions.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Build and submit a pipeline in the Hugging Face interface

Create a pipeline from the sample XGBoost workflow or from scratch, then submit it for execution and monitor the run.

Before you start

  • A Hugging Face account with a Pro subscription is required to run a pipeline.
  • Login is required to submit a pipeline.

Steps

  1. Open Tangle at Hugging Face and choose the sample XGBoost pipeline or start a new one.

  2. Drag components onto the canvas, connect outputs to inputs, and configure task arguments.

  3. Submit the pipeline for execution.

  4. Monitor the run in the interface.

Expected outcome

A submitted pipeline run that can be monitored in Tangle.

Guide limitations

  • You can build without registration, but running a pipeline requires Hugging Face login and a Pro subscription.
  • Detailed job configuration is not provided in the setup instructions.
Documentation 1 ↗Documentation 2 ↗
Source-based practical guide

Set up Tangle locally

Install and open the local Tangle app so you can begin building pipelines.

Before you start

  • Docker and uv installed.
  • A terminal and web browser.

Steps

  1. Clone the Tangle backend repository using the stable branch and the Tangle UI repository using the stable_local_build branch.

  2. On Linux or macOS, run cd tangle && backend/start_local.sh; on Windows, run cd tangle && backend\start_local.cmd.

  3. When the terminal reports that the orchestrator is starting, open http://localhost:8000 in a browser.

  4. Choose New Pipeline to begin building.

Expected outcome

The local Tangle app opens in a browser and is ready for pipeline creation.

Guide limitations

  • The steps cover local setup and opening the app, not execution configuration for a particular pipeline.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗Guide documentation ↗
Source record details
Source availability
Active
Sources checked
4 October 2026
Original website title
TangleML/tangle · Official homepage title · retained in its source language
How we review sourcesSuggest a correction

Related options

Kubeflow Pipelines — Tangle uses the ComponentSpec format associated with Kubeflow Pipelines, allowing components to be reused between the systems.

Vertex AI Pipelines — Tangle supports reuse of Vertex AI pipeline components.

Apache Airflow — Tangle’s documentation contrasts its visual, containerized ML pipeline model and execution caching with Airflow’s code-defined workflows.