Learn by teaching
Can you explain it?
The real test of understanding is explaining it to someone who knows nothing. Sam just started at the coffee chain's office and has never seen a database. Explain a topic in plain words; Sam asks questions; then you get a verdict on what came across and what didn't.
Part of Pro
Pro includes 20 teach-backs a month, one topic for every lesson you finish.
Create a free account, then see Pro.
What a data pipeline is
Explain to Sam what a data pipeline is and why the coffee chain needs one.
Unlocks when you finish its lesson.
Rows, columns and keys
Explain to Sam what a key column is, and why a table needs one.
Unlocks when you finish its lesson.
Filtering with WHERE
Explain to Sam how to get only the Austin orders out of a table, and what can go wrong.
Unlocks when you finish its lesson.
Grain and GROUP BY
Explain to Sam what GROUP BY does, and what "grain" means.
Unlocks when you finish its lesson.
Raw data, clean data
Explain to Sam why we keep the raw data and build a separate clean table.
Unlocks when you finish its lesson.
Idempotent loads
Explain to Sam what an idempotent load is and why pipelines need them.
Unlocks when you finish its lesson.
JOIN and LEFT JOIN
Explain to Sam what a JOIN is, and when you would use a LEFT JOIN instead.
Unlocks when you finish its lesson.
Fan-out
Explain to Sam why joining orders to order items made the delivery-fee total too big, and how to fix it.
Unlocks when you finish its lesson.
What git actually is
Explain to Sam what git is, beyond "a backup".
Unlocks when you finish its lesson.
Staging and history
Explain to Sam the difference between the working files, the staging area and a commit, and how to go back in time.
Unlocks when you finish its lesson.
Branches
Explain to Sam what a branch really is in git, and what merging does.
Unlocks when you finish its lesson.
Merge conflicts
Explain to Sam what a merge conflict is and how to resolve one calmly.
Unlocks when you finish its lesson.
The command line for data
Explain to Sam how to look inside a huge CSV and count rows per city without opening it.
Unlocks when you finish its lesson.
Debugging a failed pipeline
Explain to Sam how you would investigate a pipeline that failed overnight.
Unlocks when you finish its lesson.
Reading someone else's SQL
Explain to Sam how to find a bug in a long SQL model they didn't write.
Unlocks when you finish its lesson.
Images and containers
Explain to Sam what Docker images and containers are, and why data teams use them.
Unlocks when you finish its lesson.
Query plans
Explain to Sam what a query plan is and how it helps with a slow query.
Unlocks when you finish its lesson.
Cleaning messy text values
Explain to Sam why one shop can show up as several cities in a report, and how a pipeline fixes it for good.
Unlocks when you finish its lesson.
Data quality tests
Explain to Sam what a data quality test is, what kinds there are, and what should happen when one fails.
Unlocks when you finish its lesson.
Latest wins with window functions
Explain to Sam how to keep only the newest version of each customer when the app sends a new row for every profile change.
Unlocks when you finish its lesson.
Incremental loads
Explain to Sam how an incremental load works, and how it can lose rows without any error.
Unlocks when you finish its lesson.
Orchestration, retries and backfills
Explain to Sam what an orchestrator does, and why retries and backfills only work if every task is idempotent.
Unlocks when you finish its lesson.
Facts and dimensions
Explain to Sam why the wide orders export gets split into a fact table and dimension tables, and what goes where.
Unlocks when you finish its lesson.
Star schemas and surrogate keys
Explain to Sam what a star schema is, and why its dimensions use surrogate keys and an unknown member.
Unlocks when you finish its lesson.
Slowly changing dimensions
Explain to Sam why overwriting a customer's city changed last quarter's report, and how Type 2 keeps history.
Unlocks when you finish its lesson.
Batch and streaming
Explain to Sam the difference between batch and streaming, and what makes streaming hard.
Unlocks when you finish its lesson.
Warehouses, lakes and lakehouses
Explain to Sam what a data warehouse, a data lake and a lakehouse are, and how to choose.
Unlocks when you finish its lesson.
Performance and cost
Explain to Sam how a cloud warehouse bill is driven by queries, and how to bring it down safely.
Unlocks when you finish its lesson.
Landing raw data
Explain to Sam how you load the tills' files into raw tables, and why you don't clean anything yet.
Unlocks when you finish its lesson.
From raw to a tested star schema
Explain to Sam how the raw orders become trustworthy fact and dimension tables.
Unlocks when you finish its lesson.
A pipeline that runs itself
Explain to Sam how the pipeline keeps itself correct every night when files arrive late, a test fails, or a job gets retried.
Unlocks when you finish its lesson.
Serving numbers you can trust
Explain to Sam how you turned the model into the owner's dashboard, and how you know its numbers are right.
Unlocks when you finish its lesson.
Loading CSV and JSON files
Explain to Sam why you can't just trust a CSV file to load correctly, and how you load one safely.
Unlocks when you finish its lesson.
Parquet and columnar storage
Explain to Sam why the data team stores files as Parquet instead of CSV.
Unlocks when you finish its lesson.
Partitioning data
Explain to Sam what partitioning is and how you would choose the partition key for the coffee chain's orders.
Unlocks when you finish its lesson.
Stacking batches with UNION
Explain to Sam how to combine the tills' daily order files into one table, and what can go wrong.
Unlocks when you finish its lesson.