metaflow-netflixext

What it is and what it does

Metaflow-netflixext is a collection of experimental extensions developed at Netflix that extend Metaflow's core capabilities beyond what is in the open-source package. The primary feature is Conda V2, a refactored environment decorator that improves upon Metaflow's standard Conda support by allowing mixed Conda and PyPI package specifications, supporting multiple PyPI sources, and providing a command-line tool to inspect, resolve, and re-hydrate environments. It also includes improved debugging capabilities and Hugging Face authentication management.

The package is positioned as a proving ground for functionality that may eventually be integrated into the main Metaflow release but is not yet ready for community inclusion—either because its utility is unclear, community resources are limited, or it requires further testing. While Netflix maintains it actively, it does not carry the same backward-compatibility guarantees as the core Metaflow package. It is intended for users who need Netflix's infrastructure integrations or want to experiment with cutting-edge Metaflow features.

Use it for:

• Build Metaflow workflows that require both Conda packages and PyPI packages in the same environment without manual workarounds.
• Inspect and re-hydrate previously executed step environments to debug artifacts or reproduce computations.
• Authenticate to Hugging Face within Metaflow workflows without managing credentials manually.
• Resolve complex mixed-dependency environments using conda-lock for reproducibility across machines.
• Cache and reuse named environments across multiple workflow runs to reduce setup overhead.

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

Provides Netflix-developed extensions to Metaflow, including an improved Conda environment decorator that mixes Conda and PyPI packages, debugging tools, and Hugging Face authentication support.

Yes, if you are already using Metaflow and need the specific features it provides—particularly mixed Conda/PyPI environment management or Hugging Face integration. The low install friction, active maintenance, and lack of known vulnerabilities make it safe to adopt. However, be aware that this is experimental functionality without the stability guarantees of core Metaflow; features may change or be removed if merged upstream. Not necessary if you only use standard Metaflow capabilities.

Install

metaflow-netflixext on PyPI

pip install metaflow-netflixext

uv add metaflow-netflixext

poetry add metaflow-netflixext

Installing metaflow-netflixext

Before you install

Low install friction with a single runtime dependency on metaflow. Active maintenance with recent releases; requires Metaflow v2.8.3 or later and Python 3.10+. Repository shows steady activity and no known vulnerabilities.

License in practice

Licensed under Apache Software License (permissive). No restrictions on commercial or private use; modifications and redistribution are permitted under standard Apache terms.

Quickstart

pip install metaflow-netflixext

from metaflow_netflixext.netflix_ext.conda_v2 import CondaV2

@CondaV2(packages={'package': 'version'})
def my_step(self):
    pass

Requires Metaflow v2.8.3 or later and Python 3.10 or higher. Local Conda, Mamba, or Micromamba installation needed for environment resolution.

Verify before relying

• Whether the Conda V2 decorator's performance improvements are quantifiable or measurable in typical workflows.
• Current stability and backward-compatibility guarantees for the Hugging Face auth management feature.
• Whether functionality from this extension has been merged into the main Metaflow package since version 1.3.16.

Package facts

Evidence: metaflow_netflixext-1.3.16-py2.py3-none-any.whl

Tags

More Distributed Computing packages