Skip to content
LakeBench
ProblemsCommunityPricing
Sign inStart practicing
Back
  1. Home
  2. Interview prep
  3. End-to-end mental model

Data platform · Platform Concepts

End-to-end mental model

Hardplatform-25
architectureend-to-endplatforminterview

Question

Walk through a modern analytics platform from source to dashboard.

Solution

Interviewers often ask for an end-to-end story. Keep it layered and name responsibilities.

1. Sources      OLTP DBs, SaaS APIs, app events
2. Ingest       Fivetran / Airbyte / CDC (Debezium) / Kafka
3. Raw storage  lake bronze or warehouse raw schema
4. Transform    dbt / Spark (silver -> gold / medallion)
5. Orchestrate  Airflow / Dagster schedules + retries
6. Quality      tests, freshness SLA, contracts, observability
7. Serve        BI dashboards, reverse ETL, ML features

Tiny concrete flow

Checkout Postgres → Debezium → Kafka orders → land to bronze.orders → dbt stg_orders → fct_orders → Looker dashboard. Airflow triggers dbt after lag/freshness gates. Unique tests protect order_id.

What "good" sounds like

You mention ownership, idempotent loads, backfills, cost controls, and who gets paged when the executive GMV dashboard is stale.

Interview tip: Draw the boxes top to bottom, then dive deep only where the interviewer pushes (Kafka, dbt, or modeling).

Previous