Python. SQL. Spark. Power BI.
One person. A whole toolkit.
4cases live
13–16 minper case
0setup required
Freeto read, always
The stack you'll touch in Season 1
Python
Apache Spark
Apache Kafka
Snowflake
Apache Airflow
PostgreSQL
Databricks
BigQuery
pandas
MongoDB
Apache Flink
ClickHouse
DuckDB
MySQL
Airbyte
Looker
Metabase
Apache Superset
Grafana- S3Amazon S3
- SQLSQL
- BIPower BI
Season 1 · Case files
Four cases live. One growing company.
The company scales from a thousand orders a day to millions. Your pipeline grows with it, and so do the ways it breaks. Start with Case 01, then work through the rest in order.
Case 01Storage
Two databases, opposite shapes
Why the database that runs your app is terrible at answering questions about it — shown byte by byte.
Case 02Batch & streaming
The two clocks
Watch a late order break a live dashboard, and the batch-vs-streaming argument changes shape.
Case 03Data quality
The green run
The job finished. Nothing errored. The numbers are wrong. Stop it before anyone sees it.
Case 04Modelling
The customer who moved
One customer moves house, and ₹2,000 of last year’s revenue moves with her.
Case 05Lineage
Where did this number come from?
Follow one wrong figure backwards through every table it touched, in minutes rather than all day.
Drops soon · Notify meCase 06Performance
The 10x Saturday
Traffic grows overnight. The job that took 4 minutes now takes 9 hours.
Drops soon · Notify meGet paged
Get paged when Case 05 drops.
Cases 01 to 04 are live now. Leave your email and we’ll send one short note the day each new case goes live. No spam, unsubscribe in a click.
How it works
You don't read a lesson. You work a case.
Every episode drops you into a real pipeline failure, the kind that wakes engineers at 3 AM. You investigate, make the call, and walk away with the concept and the fix.
- 01
The incident
A dashboard shows zero revenue on a Friday night. Something upstream broke. The clock is running.
- 02
The investigation
Trace the data backwards through serving, warehouse, transformation and ingestion. Pick your suspect.
- 03
The playbook
The reveal explains the concept behind the failure, and you keep a one-page playbook you'll use at work.
Who it's for
Anyone who wants data engineering to finally click.
Students and switchers
You've heard Kafka, Spark and Airflow a hundred times. Here you'll see where each one actually sits.
Analysts moving up
You know SQL. Now learn the pipelines behind the tables you query, and the failures nobody tells you about.
Working engineers
The incidents are real. Use them as a playbook for postmortems, design reviews and interviews.
The platform
Stories first. Then the hands-on part.
Episodes teach you how to think. The modules make you practise, in the browser, with no setup.
revenue · today₹0
StoriesCinematic episodes built on real incidents, with sources for every number.
LiveSELECT city, count(*) AS orders
FROM deliveries
GROUP BY city;Interview-grade SQL problems with instant feedback and query plans explained in plain English.
In builddf.groupBy("city")
.agg(F.count("*"))Write Spark in the browser, watch the shuffle happen, and see why the slow version is slow.
In buildWhy is my Spark job slow?One key holds 40% of the rows. That’s skew. Try salting the key.
AI TutorAsk anything mid-episode. It knows the case you're on and answers with your own numbers.
In buildDo I need to know how to code?
No. The cases don’t make you write code. Basic SQL helps you get more out of them, and the SQL module is built to get you there.
Is it free?
The episodes are free to read. The hands-on modules will have a paid tier when they launch.
How often do episodes drop?
One per week during season 1. If you leave your email, we ping you when the next case is live on the site.
Are the incidents real?
They're based on failures that happen in real production systems, rewritten with fictional companies. Every statistic carries its source.
Does it work on my phone?
Yes. The episodes are built mobile-first, with headphones recommended for the sound design.
Still deciding?