---
id: ob-metaflow
version: "2.19.37.1"
license: Apache License 2.0
license_treatment: permissive
maintenance: active
---

# ob-metaflow — Metaflow: More AI and ML, Less Engineering

License: permissive · Maintenance: active · Downloads: 113.3K/mo

## What it is and what it does

ob-metaflow is a framework for building AI and ML systems that spans the entire lifecycle from local prototyping in notebooks to reliable production deployments. It unifies code, data, and compute management, providing a Pythonic API for experiment tracking, versioning, and visualization alongside support for scaling workloads horizontally and vertically across cloud infrastructure using CPUs and GPUs.

The framework handles dependency management, data access, and orchestration, with built-in support for both embarrassingly parallel and gang-scheduled compute jobs. It integrates with cloud providers via boto3 and kubernetes, and supports deployment to production orchestrators with reactive event triggering. Originally developed at Netflix and now supported by Outerbounds, it is designed for teams ranging from individual researchers to large organizations running thousands of concurrent workflows.

Use it for:

- Prototype ML models locally in a notebook, then scale the same code to distributed cloud compute without rewriting
- Track and version experiments across multiple runs, comparing results and managing artifacts automatically
- Deploy a trained model as a production workflow that triggers on events and scales based on demand
- Run embarrassingly parallel data processing jobs across many machines with built-in failure recovery
- Manage dependencies and orchestrate multi-stage pipelines with clear data lineage between steps

## Worth the install?

AI-flagged interpretation of the facts above — verify before relying.

A framework for building, prototyping, and deploying AI and ML workflows from notebooks to production, handling data management, compute scaling, and orchestration across local and cloud environments.

Yes. ob-metaflow is actively maintained, has low install friction, carries no known vulnerabilities, and uses a permissive license. It is well-suited for teams building ML systems that need to move from prototyping to production without significant refactoring. Start with local prototyping and scale incrementally as your infrastructure needs grow.

## Install

pip install ob-metaflow
uv add ob-metaflow
poetry add ob-metaflow

## Installing ob-metaflow

Before you install: Low install friction with a pure-Python wheel distribution. Active maintenance with a release 3 days old. Runtime dependencies on requests, boto3, pylint, and kubernetes are standard infrastructure libraries.

License in practice: Apache License 2.0 is permissive, allowing commercial use, modification, and distribution with minimal restrictions—suitable for most organizational contexts.

Quickstart:

pip install ob-metaflow

from metaflow import FlowSpec, step

class MyFlow(FlowSpec):
    @step
    def start(self):
        self.next(self.end)
    
    @step
    def end(self):
        pass

if __name__ == '__main__':
    MyFlow()

Verify before relying:

- Minimum Python version requirement (requires_python is unspecified in the fact sheet)
- Whether kubernetes dependency is required or optional for local prototyping
- GPU support specifics and any system-level dependencies beyond the listed runtime packages

## Package facts

- License: Apache License 2.0 (permissive)
- Python support: unspecified
- Install friction: low
- Maintenance: active
- Downloads: 113.3K/month (top 15,000 on PyPI)
- Known vulnerabilities: none known

## Tags

ml workflow orchestration, data science pipeline framework, ml experiment tracking and versioning, distributed compute for ml, production ml deployment, notebook to production ml, ai workflow management, ml-orchestration, workflow-framework, distributed-computing

[View on SkillFed](https://skillfed.io/packages/ob-metaflow) · [View on PyPI](https://pypi.org/project/ob-metaflow/)