---
id: flytekit
version: "1.16.27"
license: Apache-2.0
license_treatment: permissive
maintenance: active
---

# flytekit — Flyte SDK for Python

License: permissive · Maintenance: active · Downloads: 545.8K/mo

## What it is and what it does

Flytekit is the Python SDK for the Flyte workflow orchestration platform. It provides decorators and APIs to define tasks (atomic units of work) and workflows (compositions of tasks) with built-in support for caching, retries, and type safety. Tasks and workflows are authored in plain Python, then deployed to a Flyte backend for distributed execution. The package handles serialization, dependency management, and integration with cloud storage (GCS, Azure, S3) and container runtimes (Docker).

The library is designed for data engineers and ML practitioners building reproducible, scalable pipelines. It includes 39 runtime dependencies covering serialization (cloudpickle, msgpack, protobuf), scheduling (croniter), caching (diskcache, cachetools), and cloud integrations (gcsfs, adlfs). While Flyte 2 is now generally available, this version (1.16.27) remains actively maintained and suitable for existing Flyte 1.x deployments.

Use it for:

- Define and test ML training pipelines with task caching and automatic retry logic before deploying to a Flyte cluster.
- Build data processing workflows that orchestrate multiple Python functions across distributed infrastructure with type-safe interfaces.
- Schedule recurring data jobs using cron expressions and manage their execution state through Flyte launch plans.
- Integrate with cloud storage backends (GCS, Azure Blob, S3) for reading and writing pipeline artifacts.
- Author reusable task libraries with built-in versioning and dependency tracking for team collaboration.

## Worth the install?

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

Flytekit is the Python SDK for authoring, testing, deploying, and managing Flyte tasks, workflows, and launch plans—a framework for building data and ML pipelines that run on the Flyte backend.

Yes, if you are running Flyte 1.x or evaluating Flyte for workflow orchestration. The package is actively maintained, has low install friction, carries no known vulnerabilities, and is permissively licensed. However, note that Flyte 2 is now generally available—new projects should evaluate whether Flyte 2's Python SDK is a better fit, and existing Flyte 1.x users should plan a migration timeline.

## Install

pip install flytekit
uv add flytekit
poetry add flytekit

## Installing flytekit

Before you install: Low install friction with a pure-wheel distribution. Actively maintained with a release 4 days old and recent commits; 315 repository stars indicate established community use.

License in practice: Apache-2.0 permissive license allows commercial use, modification, and distribution with minimal restrictions—suitable for most production and proprietary contexts.

Quickstart:

pip install flytekit

from flytekit import task, workflow

@task(cache=True, cache_version="1", retries=3)
def sum(x: int, y: int) -> int:
    return x + y

@workflow
def my_workflow(x: int, y: int) -> int:
    return sum(x=x, y=y)

Requires Python 3.10 or later (capped below 3.13); Flyte backend services must be available or configured for full functionality.

Verify before relying:

- Whether Flyte 1.x receives ongoing security patches or is in maintenance-only mode given Flyte 2 general availability.
- Performance characteristics and scalability limits for large-scale workflow execution.
- Compatibility and migration path from Flyte 1.x to Flyte 2 for existing deployments.

## Package facts

- License: Apache-2.0 (permissive)
- Python support: capped_below_current
- Install friction: low
- Maintenance: active
- Downloads: 545.8K/month (top 15,000 on PyPI)
- Known vulnerabilities: none known

## Tags

python workflow orchestration, data pipeline framework, flyte sdk python, ml workflow management, task scheduling and caching, distributed workflow execution, flyte task decorator, workflow-orchestration, ml-pipeline, distributed-computing

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