Quickstart

Let’s get you up and running with your first workflow on your local machine.

What you’ll need

• Python 3.10+ in a virtual environment

Install the SDK

Install the flyte package:

pip install 'flyte[tui]'

Verify it worked:

flyte --version

Output:

Flyte SDK version: 2.*.*

uvx flyte --version
uvx flyte get run

Run something straight away

Before writing anything of your own, you can run a built-in example. It needs no files and no configuration:

flyte run --local hello

Flyte writes the example to a scratch directory, runs it, and prints the path to the source:

Using the built-in example from /tmp/flyte-hello-<user>/task/hello.py
Copy it into your own project to start editing.
Completed Local Run   Outputs: ActionOutputs(o0=14.0)

The example fans a small computation over a list of inputs with flyte.map and averages the results. That is enough to see a workflow run. Next, write one of your own.

flyte run --tracked hello

Configure

Create a config file for local execution. Runs will be persisted locally in a SQLite database.

flyte create config --local-persistence

This creates .flyte/config.yaml in your current directory.

Write your first workflow

This one converts a list of temperature readings and returns the hottest. Create temperatures.py:

# temperatures.py

import flyte

# A TaskEnvironment groups configuration for the tasks defined within it:
# the container image, resources, and so on. This one keeps the defaults.
env = flyte.TaskEnvironment(name="temperatures")

# The @env.task decorator turns a Python function into a task.
# Type annotations on the inputs and output are required.
@env.task
def to_fahrenheit(celsius: float) -> float:
    return celsius * 9 / 5 + 32

# This is the entrypoint task of the workflow. It calls to_fahrenheit
# once per reading using flyte.map, which is like Python's map but runs
# the calls in parallel, then returns the highest result.
@env.task
def hottest(readings: list[float] = [21.5, 19.0, 24.3, 22.8]) -> float:
    return round(max(flyte.map(to_fahrenheit, readings)), 1)

Here’s what’s happening:

• TaskEnvironment specifies configuration for your tasks (container image, resources, etc.)
• @env.task turns Python functions into tasks that can run on a cluster
• flyte.map calls to_fahrenheit once per reading, in parallel when running on a cluster
• Both tasks share the same env, so they’ll have identical configurations

Run it

Create a project directory and place your files there:

.
├── temperatures.py
└── .flyte
    └── config.yaml

Run the workflow, naming the file and the entrypoint task:

flyte run --local temperatures.py hottest

This executes the workflow locally on your machine:

Completed Local Run
Outputs: ActionOutputs(o0=75.7)

See the results

You can see the run in the TUI by running:

flyte start tui

The TUI will open into the explorer view

To navigate to the run details, double-click it or press Enter to view the run details.

Next steps

Now that you’ve run your first workflow: