The following is a very simple example of an Orca pipeline with none of the optional features such as caching:

import orca
import pandas as pd

def data_file():
    return 'data.csv'

def raw_data(data_file):
    return pd.read_csv(data_file)

def processed(raw_data):
    # do fancy stuff
    return processed_data

def save_data(processed, data_file):
    processed.to_csv('processed_' + data_file)

def save_fig(processed):
    # save fancy figure

orca.run(['save_data', 'save_fig'])

By declaring the data dependencies of your functions you can let Orca thread them all together instead of doing it manually yourself.

The iterative capabilities of Orca are illustrated by imagining that there are many data files to be processed:

def data_file(iter_var):
    return iter_var

# everything else the same as above

files = glob.glob('*.csv')
orca.run(['save_data', 'save_fig'], iter_vars=files)

A more involved example is available at https://gist.github.com/jiffyclub/2a252333c8dcad1b99aa.