about d8loop

Most practice tests what you remember.D8LooP tests what you’d decide.

Most data engineering practice teaches you to recall. The job asks you to decide.

You can memorise every join type and every partitioning strategy and still freeze when someone asks why this grain and not that one. Recall is testable, so that’s what practice sites test. But nobody gets stuck at work because they forgot a syntax rule. They get stuck because a table has one mutable status column and the question on the table is what the status was last Tuesday.

D8LooP exists for that second thing. You design something, it gets picked apart, and you have to defend the choice you made.

Marketplace core entities·Fulfillment network·Creator payout operations·City parking bays·Food delivery orders·Cinema seat booking·Social feed and follows·Dating app matches·Ride hailing trips·Category performance over time·Seat holds and the release-night race·Matches at scale·Watch time by title·Campaign efficiency·Churn that survives an argument·Which courses actually work·Engagement from a messy event log·Orders you can trust eight months later·Dispatch that can't double-book a driver·A ledger a regulator will accept·Deletion that doesn't break the thread·Stock you can explain·Org chart and project staffing·Refunds without double counting·Live dispatch and daily KPIs·Current plan and monthly recurring revenue·Addresses that stay true to the past·A hot dashboard over cold history·Enrollment state and honest completion rates·Subscription warehouse grain·Support ticket lifecycle·Product analytics identity·Loads, drivers, and backhauls that vanish·Patients, sites, and protocol amendments mid-study·Interval readings, estimates, and bill corrections·Claims, reserves, and payments that reverse·Events, tickets, and transfers that can be undone·Vessels, berths, and overlapping time windows·Loans, drawdowns, and partial prepayments·Fields, sensors, and missing readings·Releases, rights holders, and territory splits·Impressions, bids, and retroactive disqualifications·Care episodes, providers, and overlapping responsibilities·User actions, sessions, and retroactive merges·Inventory, shipments, and in-flight adjustments·Players, currencies, and promotional grants·Jobs, technicians, and skill hierarchies·Policies, endorsements, and mid-term changes·Datasets, owners, and access grants·Trades, currencies, and rates as of trade date·Routes, bins, and missed pickups·Titles, rights holders, and format windows·Customer 360 tradeoffs·Global pricing history·Risk and compliance audit·Subscriptions & billing·Marketplace core entities·Fulfillment network·Creator payout operations·City parking bays·Food delivery orders·Cinema seat booking·Social feed and follows·Dating app matches·Ride hailing trips·Category performance over time·Seat holds and the release-night race·Matches at scale·Watch time by title·Campaign efficiency·Churn that survives an argument·Which courses actually work·Engagement from a messy event log·Orders you can trust eight months later·Dispatch that can't double-book a driver·A ledger a regulator will accept·Deletion that doesn't break the thread·Stock you can explain·Org chart and project staffing·Refunds without double counting·Live dispatch and daily KPIs·Current plan and monthly recurring revenue·Addresses that stay true to the past·A hot dashboard over cold history·Enrollment state and honest completion rates·Subscription warehouse grain·Support ticket lifecycle·Product analytics identity·Loads, drivers, and backhauls that vanish·Patients, sites, and protocol amendments mid-study·Interval readings, estimates, and bill corrections·Claims, reserves, and payments that reverse·Events, tickets, and transfers that can be undone·Vessels, berths, and overlapping time windows·Loans, drawdowns, and partial prepayments·Fields, sensors, and missing readings·Releases, rights holders, and territory splits·Impressions, bids, and retroactive disqualifications·Care episodes, providers, and overlapping responsibilities·User actions, sessions, and retroactive merges·Inventory, shipments, and in-flight adjustments·Players, currencies, and promotional grants·Jobs, technicians, and skill hierarchies·Policies, endorsements, and mid-term changes·Datasets, owners, and access grants·Trades, currencies, and rates as of trade date·Routes, bins, and missed pickups·Titles, rights holders, and format windows·Customer 360 tradeoffs·Global pricing history·Risk and compliance audit·Subscriptions & billing·

why it exists

Nobody gets hired for reciting a definition. Turn any one of these over.

how practice works here

Three beats, every scenario — and each one runs on its own here.

Build

Design the pipeline or the schema. You get the business problem — never the table names.

Review

Scored on structure, requirements, and how it fails. Not against one blessed answer.

Defend

Then the probes. What if it arrives twice? What was this value last Tuesday?

who it's for

Interviewing soon
You can write the query. You have never been asked to justify the schema under it.
Two years in
Everything you have built started from a table someone else designed.
Senior, out of practice
You make these calls weekly. Saying why, in nine minutes, is the rusty part.

who's behind it

Vijay Singh

2019 — 2026AmazonData Engineer
2026 — nowDunnhumbySenior Big Data Engineer

I have been 7 years at Amazon, on both sides of the interview table. I built the thing I kept wishing candidates had.

Email LinkedIn

Most practice tests what you remember.D8LooP tests what you’d decide.

Interviewing soon

Two years in

Senior, out of practice

Vijay Singh