One Course
Every lesson, project and mock exam in Apache Airflow.
Billed monthly. Cancel any time.
Level 1: Newcomer
0 points
Sign in to save your progress across devices.
Courses / Data and analytics tools
Learn Apache Airflow by writing real Dags in a working Airflow workspace. Each lesson has a worked example, then tasks that run your Dag and check what happened: TaskFlow, scheduling, XCom, sensors, retries and testing.
with DAG and @dag, and tasks with the TaskFlow API and the classic operators>>, lists and chain, and fix the import errors that hide a DagUnit 1, Dags, tasks and dependencies: 6 lessons and 33 checked exercises, with no account or card.
Start the free unitThe rest of the course opens with a One Course or All Access plan. See plans and prices
Choose access here. Your course is already selected.
Build Marigold's nightly sales pipeline step by step: the right day, retries, an idempotent load, a report, alerts, a deadline and a backfill.
8 stepsProject · AdvancedBramble's morning parcel trackerBuild Bramble's morning tracker: map over the depots that reported, track every open parcel within the partner's limits, save the results and start the customer texts.
7 stepsCase study · ExpertA week stuck behind one bad dayFind why a daily ledger stopped on Thursday, unblock the week safely, and make sure the next failure is noticed on the day.
5 stepsCase study · ExpertThe upgrade that moved the dayTrace a wrong date after the Airflow 3 upgrade to the new cron timetable, fix it without touching task code, and backfill the history.
5 stepsCase study · ExpertThe morning the scanners stalledExplain why an on-time depot waits behind late ones, free the slots, stop waiting forever on a depot that may not upload, and recommend a design.
5 stepsProject · ExpertFenwick Pay's settlement reconciliationBuild Fenwick Pay's nightly reconciliation: map over merchants, store results, alert on mismatches, wait for the network's file, schedule it by business day and hand the results on as an asset.
7 stepsMock exam · ExpertAirflow job test A: fundamentals (timed)A timed, 20-question Airflow test on Dags, scheduling, runs, the UI and CLI, XCom, variables, connections and debugging.
20 questions · 40 minMock exam · ExpertAirflow job test B: Dag authoring (timed)A timed, 20-question Airflow test on TaskFlow, templating, branching, trigger rules, sensors, mapping, assets, testing and pools.
20 questions · 45 minPractice Apache Airflow in a browser workspace that parses your Dag files and runs them against a simulation of Airflow 3.3: the Dags list, Grid and Graph views, task logs, XComs, variables, connections, pools, assets and a terminal with the airflow command line. Start with your first Dag and the TaskFlow API, then schedule it, pass data between tasks, retry what fails, wait for files with sensors, branch, map tasks over lists and trigger Dags from assets. Later levels cover backfills, testing, production patterns and incidents, with projects, interview sets and timed job tests. Every task is checked on your data and on a hidden copy. No installation is needed.
Explain what you've learned in a full-screen interview. Speak or type, then review your answers against the rubric for your level.
Interview loops by jobBeginner
Write Dag files with DAG and @dag, pass data with TaskFlow, set dependencies and use the classic operators.
Trigger and pause runs, read task logs, use the airflow command line and fix import errors.
Pass values between tasks, give Dags inputs, read settings and credentials, and render templates.
Intermediate
Choose schedules, understand logical dates and data intervals in Airflow 3, replay history with catchup and backfills, and schedule in local time.
Retry what is transient, stop what hangs, alert on failures and late runs, make loads idempotent and rerun tasks safely, then build Marigold's nightly pipeline.
Advanced
Decide when tasks run with trigger rules, choose paths with branches and short circuits, and organize tasks with groups, setup and teardown.
Wait for files, reports and other Dags without wasting worker slots, and link Dags that depend on each other.
Start Dags when the data they read is updated, combine assets with conditions and clocks, pass details in asset events, and write assets with @asset.
Fan out over lists known only at run time, pair and fix arguments, protect rate-limited systems, then build Bramble's morning parcel tracker.
Expert
Catch broken files with DagBag tests, test task logic with pytest, and run whole Dags in CI with dag.test().
Share systems safely with pools and priorities, know the Airflow 3 components, write Dag files that are cheap to parse and safe to rerun, and compare Airflow with Dagster.
Work through realistic incidents from symptom to fix to the summary for the business, then build Fenwick Pay's settlement reconciliation end to end.
Practice Airflow interview questions at three levels and take two timed job tests on fundamentals and Dag authoring.
Every lesson builds on files like this one: a Dag defined with @dag, tasks written as Python functions with @task, and the date of the run read from Airflow rather than typed in.
import pendulum
from airflow.sdk import dag, task
@dag(schedule="@daily", start_date=pendulum.datetime(2026, 11, 1, tz="UTC"))
def marigold_daily():
@task
def extract_sales(ds=None):
return fetch_sales(ds)
@task
def total_revenue(rows):
return round(sum(r["revenue"] for r in rows), 2)
total_revenue(extract_sales())
marigold_daily()
Practice Apache Kafka to inspect keyed records, partition ordering and consumer offsets before querying downstream data. Use the Kafka command reference to connect each console action with the state it changes.
No. The workspace runs a simulation of Airflow 3.3 in your browser: it parses your Dag files, runs them on a simulated scheduler and shows the Dags list, Grid, Graph, logs and a terminal with the airflow command line. Where it simplifies, the lesson says so.
Airflow 3, with the Task SDK imports (from airflow.sdk import dag, task), schedule instead of schedule_interval, and the Airflow 3 defaults for catchup and timetables. Changes in 3.4 are named where they matter.
The first unit is free and needs no account. The rest of the course opens with a One Course or All Access plan, and the pricing page shows every plan and its price.
By running it: the check parses your Dag file, runs the Dag and looks at the result (task states, XCom values, rendered commands, loaded rows), on your data and on a hidden copy. A wrong answer gets a hint about what probably went wrong.
No. The Work Gym is independent and not affiliated with the Apache Software Foundation. Apache, Apache Airflow and Airflow are trademarks of the Apache Software Foundation.