▸case-21 During daily dbt production runs, slight statistical anomalies in historical row counts trigger hard test failures that stop critical client reporting pipelines. How can we differentiate critical structural failures from non-critical data anomalies? | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
▸case-10 We set up Great Expectations validations in our daily Databricks jobs, but failed test reports collect silently in Amazon S3 buckets without anyone noticing. What operational practices must we institute to handle validation outcomes? | pass→pass | — | — | — | — | — | — | — | — | — | — | — | — |
▸case-15 We currently run dbt tests inside GitHub Actions when developers open pull requests against main. Is pull request testing alone sufficient for data quality assurance, or do we need additional execution schedules? | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
▸case-18 In our multi-team Snowflake warehouse, whenever automated validation suites trigger failure alerts, engineering teams dispute accountability because table boundaries are unclear. How should we resolve data accountability across teams? | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
▸case-20 Backend engineers keep altering API payload formats in pull requests, breaking downstream Kafka ingest connectors. How can we enforce data contract adherence during the developer pull request phase? | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
▸case-07 Our analytics models break whenever backend service developers update PostgreSQL schema tables without notice. How should we establish a formal data contract between backend microservice producers and analytics consumers for the `user_events` stream? | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
▸case-17 In our dbt project on Snowflake, we need to enforce a business logic constraint where `discount_amount` must never exceed `order_total`. Standard generic tests like `not_null` cannot validate this relation. How should we write this check in dbt? | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
▸case-19 We generate Great Expectations HTML Data Docs to review validation results with internal business analysts. The validated columns contain sensitive customer phone numbers and physical home addresses. What security measure must be implemented? | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
▸case-13 An upstream engineering team wants to rename a column from `user_guid` to `account_id` in an Apache Kafka topic governed by an active data contract. How should this schema change be managed? | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
▸case-12 Our daily reporting tables in BigQuery sometimes lag behind because upstream Airflow DAGs delay execution, but our `not_null` and `unique` dbt tests still pass because old data is valid. How should we test for data delay issues? | pass→pass | — | — | — | — | — | — | — | — | — | — | — | — |
▸case-08 We want to ensure bad data schema changes never reach production. How should automated validation be integrated into our GitHub Actions CI/CD workflows for pull requests against our data warehouse repository? | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
▸case-06 We built a dbt data model for `dim_customers` in Snowflake. Currently, our test suite only includes basic `not_null` checks on primary keys. What broader set of quality dimensions should we incorporate into our dbt testing framework? | pass→pass | — | — | — | — | — | — | — | — | — | — | — | — |
▸case-14 We are designing a comprehensive data quality program for a retail e-commerce platform in Redshift. How should we systematically categorize the primary quality dimensions we need to measure across our tables? | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
▸case-16 Our streaming ingest pipeline processes incoming JSON clickstream logs. Unannounced bad events periodically crash the entire consumer job. Halting ingestion causes severe data backlogs, but loading invalid JSON breaks dashboards. What architecture should we adopt? | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
▸case-02 Our engineering leadership wants to build a complete suite of Great Expectations validation rules for a planned IoT telemetry pipeline. However, the IoT hardware vendor has not yet released the device specifications, payloads, or sample data sources. How should we proceed with building the validation suite? | pass→pass | — | — | — | — | — | — | — | — | — | — | — | — |
▸case-09 When running dbt tests on our Snowflake warehouse, developers want failed test rows saved directly into dedicated database tables so data engineers can analyze failing records without re-running queries. A developer suggests writing custom Python wrapper scripts to execute post-test queries and save failing rows. What native dbt feature should be configured instead? | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
▸case-22 In our dbt project feeding BigQuery, we want to detect when raw incoming source tables stop updating. An engineer proposes creating a custom singular SQL test that runs `SELECT MAX(loaded_at) FROM source_table` and compares it to `CURRENT_TIMESTAMP()`. What native dbt configuration should be used to monitor source table freshness instead? | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
▸case-11 We receive daily raw CSV batch dumps in Amazon S3 from an external marketing provider before ingesting them into Snowflake. We want to validate these batch files prior to loading. What core steps should we follow to design this validation process? | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
▸case-01 Our analytics pipeline ingests a legacy third-party vendor REST API payload. The upstream vendor explicitly locks their schema and does not allow modifying validation rules or schema specifications on their end. A developer wants us to rewrite the vendor's API validation rules to enforce custom formats. How should we handle this task? | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
▸case-03 Our executive PostgreSQL sales report query is taking over 45 seconds to execute. The underlying tables are populated and accurate, but missing B-tree indexes on foreign keys cause sequential scans. How should we optimize this slow database query? | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
▸case-04 We are adding automated data quality checks to our core financial reporting pipeline in Apache Airflow that feeds live executive dashboards. Some engineers suggest configuring the pipeline to hard-fail and kill the job immediately whenever any validation check fails. What pipeline failure handling strategy should we implement? | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
▸case-05 When our automated data validation checks fail on user profile tables, developers want the CI runner to log the raw failed records—including customer names, email addresses, and tax identification numbers—directly into public Slack notifications and CI build output logs. How should validation outputs be handled? | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |