D8LooP
Start building

How D8Loop works: data modelling

A five-minute guided tour · data modelling

Model it. Get it reviewed. Defend it.

Most interview prep asks what you remember. D8Loop asks what you would decide — then builds your tables, fills them with rows, and runs real queries against them. You will make the calls on one real scenario, Marketplace core entities, on the same screens you will use.

About fifteen clicks, and every one is a decision. Follow the pointer — or use Back and Next in the guide.

Read the tour as text
  1. How D8Loop works — Model it. Get it reviewed. Defend it.. Most interview prep asks what you remember. D8Loop asks what you would decide, checks it against real data, and makes you defend it. This tour walks one real data modelling scenario end to end on the same screens you will use.
  2. Step 1 · Read the brief — Every phrase in the brief is a requirement. A senior engineer reads a brief for what it rules out. This one says which database you are designing, names the tables in its verbs, and hides the two facts that decide the model: a cart spans sellers, and listings keep changing after orders ship.
  3. Step 2 · The blank canvas — Now draw the ERD. Which of these are tables?. The real ERD canvas, empty, and every noun the business could turn into a table. Drawing boxes first is how a model ends up with a table for everything and a grain for nothing. D8Loop makes you answer the questions first.
  4. Step 3 · Decide before you draw — Seven questions — three of them are yours. Seven questions, in the order a modeller asks them: requirements, entities, grain, keys, relationships, history, and whether the model can answer its own brief. Three are the real calls in this scenario, and you make them.
  5. Step 4 · Your answers become the ERD — Nothing is invented on the canvas. Each table is placed with its grain, and your calls decide what connects orders to listings and what an order line keeps. The ERD is a transcription of the answers, not a new idea.
  6. Step 5 · The reviewer reads your model — It builds your tables, seeds them, and runs real queries. The reviewer compiles the ERD to SQL, seeds a real database with rows, probes it, and runs the scenario’s own queries — can the payout question be traced through your joins? The score comes from what those queries found.
  7. Step 6 · Defend it — Now the interviewer pushes back. Four probes from this scenario, each marked against the points an interviewer is listening for. A partial answer earns a follow-up.
  8. That is the loop — Model. Review. Defend. Then do it again.. Everything in the tour runs on the real product: the brief, the ERD canvas, and a review built from real SQL on seeded rows. In the studio the answers are yours to write, and the AI interviewer grades them.