What should you review for a data modeling interview?
Where should you start?
Start with schema design. It appeared in 104 of 274 interviews, making it the largest topic in these interviews. Questions cover application databases, business workflows, industry domains, and machine learning datasets. Check whether you can explain the model you would choose, not just name its parts.
- Can you turn a business workflow into entities, relationships, keys, and constraints?
- Can you explain your choice of a relational, document, or graph model?
- For a machine learning dataset, can you define the prediction target and what each row represents?
Which repeated questions should you check first?
The most asked question was about what slowly changing dimensions are and when to use them. It appeared in 8 of 274 interviews. Choosing between star and snowflake schemas was another repeated question.
- Can you explain when you would use slowly changing dimensions?
- Can you describe how effective dates, snapshots, and late-arriving data affect historical records?
- Can you define a fact table's grain, meaning what each row represents, and identify its dimensions and measures?
- Can you explain when you would choose a star schema over a snowflake schema?
What should you know about warehouses and lakehouses?
Warehouse architecture appeared in 62 of 274 interviews. Questions cover warehouses, lakes, marts, OLTP versus OLAP, medallion layers, and migration planning. Lakehouse questions also cover Apache Iceberg and Delta Lake.
- Can you explain the difference between a data lake and a warehouse?
- Can you describe the bronze, silver, and gold layers?
- Can you explain what Iceberg and Delta Lake provide for managing tables?
- Can you explain what a catalog does and how different engines access lakehouse tables?
How would you check quality and performance?
Look beyond the initial schema. Quality and governance questions cover shared definitions, duplicate keys, source-of-truth decisions, and schema changes. Performance questions cover how data is stored and queried.
- Can you explain how you would check referential integrity and reconcile data from different sources?
- Can you explain when you would use partitioning, clustering, or indexes, and how you would read a query plan?
- For reporting, can you explain relationships, calculations, and how values are grouped and totaled in a semantic model?
Should you write SQL as part of your review?
Coding tasks appeared in 3 of 274 interviews. Check whether you can design SQL tables for a gym management system, query medication aliases, and write a SQL CTE using sales and product tables in a star schema.
Who gets asked Data Modeling
The 20 most asked Data Modeling questions
Counted by interview, not by mention: a question asked twice in one conversation counts once.
What are slowly changing dimensions, and when would you use them?
Asked in 8 of 274 interviews · last asked Oct 7 · Concept · Mid-level
Show answer
Slowly changing dimensions are warehouse dimensions whose attributes change over time, such as a customer's address. SCD methods control whether we overwrite old values or keep history. I use them when descriptive data changes, especially when reports must show what was true at the time.
- Type 1 overwrites old values. Use it for corrections or when history is not needed.
- For Type 2, match by business key, detect attribute changes, and close the old version's effective period.
- Insert a new version with a new surrogate key, effective dates, and a current-row flag.
- Link facts to the version valid when the event occurred. Avoid overlapping effective periods.
What the interviewer checks: The interviewer checks whether you can choose between overwriting values and keeping history, and correctly track Type 2 versions.
Other ways it was asked
- “How would you implement a Slowly Changing Dimension Type 2 in an ETL process?”
- “How do you implement Slowly Changing Dimension Type 2 (SCD Type 2) in a data warehouse?”
- “What are slowly changing dimensions (SCDs) in data warehousing?”
When would you use a star schema versus a snowflake schema?
Asked in 7 of 274 interviews · last asked Sep 22 · Concept · Mid-level
Show answer
Use a star schema for simple reporting, with a central fact table linked to wide dimension tables. Use a snowflake schema when splitting dimensions into related tables helps manage hierarchies and reduce repeated data. Star schemas need fewer joins; snowflake schemas add joins and query complexity. Actual query speed depends on the database and workload.
- Star schemas keep dimension attributes together; snowflake schemas normalize dimensions into related tables.
- Star schemas are usually easier for analysts and reporting tools to query.
- Snowflake schemas reduce repeated dimension values but require more joins.
- Both designs use fact tables to hold measures and links to dimensions.
What the interviewer checks: The interviewer wants to see whether you can balance reporting simplicity, repeated data, and query complexity.
Other ways it was asked
- “When would you use a star schema or a snowflake schema for data modeling?”
- “What is the difference between star and snowflake schemas, and when would you use each?”
- “What is the difference between a snowflake schema and a star schema?”
How would you design and optimize an application's database schema, and what factors would guide your decisions?
Asked in 6 of 274 interviews · last asked Oct 12 · Scenario · Mid-level
Show answer
I start with the application's data, relationships, and main queries. I choose a database based on transaction needs, data shape, expected growth, and team skills. I define tables, keys, and constraints, usually normalizing to avoid duplicate data. Then I test realistic workloads, add useful indexes, and denormalize only when measurements justify it.
- Use primary keys, foreign keys, and constraints to enforce data rules.
- Choose indexes for common queries; they use storage and slow writes.
- Check query plans and response times before changing the schema for speed.
- Use versioned migrations, access controls, backups, and restore tests to manage the database safely.
What the interviewer checks: The interviewer checks whether you connect application needs to data correctness, performance, and safe database changes.
Other ways it was asked
- “How would you design a database schema for an application, including its tables and choice of database tools?”
- “How would you design a database schema and manage the database for an application?”
- “What factors do you consider when designing a database and its tables, and how do you create them?”
What is the difference between a data lake and a data warehouse?
Asked in 5 of 274 interviews · last asked Oct 9 · Concept · Junior
Show answer
A data lake stores data in many formats, often before it is cleaned or modeled. A data warehouse stores organized data built for reporting and analysis. Lakes offer flexibility for exploration and machine learning, while warehouses make business reporting easier; modern systems can support both roles.
- Lakes can hold raw files, logs, images, and structured tables.
- Warehouses usually hold cleaned, modeled data with consistent business definitions.
- Lakes often apply structure when reading; warehouses usually define structure before loading.
- Both need data quality checks, access controls, and clear ownership.
What the interviewer checks: The interviewer is checking whether you can match storage and modeling choices to how data will be used.
Other ways it was asked
- “What is the basic difference between a data lake and a data warehouse?”
- “What is the difference between a data warehouse and a data lake?”
- “What is the difference between a database and a data warehouse?”
Can you explain database normalization and the first three normal forms with examples?
Asked in 4 of 274 interviews · last asked Sep 8 · Concept · Mid-level
Show answer
Normalization reduces repeated data and prevents update problems. 1NF stores one value per cell, with no repeating groups. 2NF also requires every non-key column to depend on each whole candidate key, not part of one. 3NF also removes dependencies where a non-key column depends on a key through another non-key column.
- Each normal form must satisfy the previous forms.
- Use candidate keys and column dependencies to decide how to split tables.
- Link the resulting tables with foreign keys.
- Normalization reduces insert, update, and delete problems, but queries may need more joins.
Code1NF: One row per product in an order. OrderLines(OrderID, ProductID, CustomerID, CustomerName, Quantity) Key: (OrderID, ProductID). OrderID determines CustomerID; CustomerID determines CustomerName. 2NF: Move customer details, which depend only on OrderID. Orders(OrderID, CustomerID, CustomerName) OrderLines(OrderID, ProductID, Quantity) 3NF: Move CustomerName, which depends on CustomerID. Orders(OrderID, CustomerID) Customers(CustomerID, CustomerName) OrderLines(OrderID, ProductID, Quantity)What the interviewer checks: The interviewer checks whether you can use keys and dependencies to split tables and avoid inconsistent data.
Other ways it was asked
- “Can you give a practical example of how database normalization is applied?”
- “What are the first three normal forms in database normalization?”
- “Can you explain the first three normal forms in database normalization?”
How would you define the prediction target and data granularity for a machine learning system?
Asked in 4 of 274 interviews · last asked Aug 31 · Design · Mid-level
Show answer
I would define exactly what the model predicts, when it predicts it, and how far ahead. I would define what one data row represents, such as one customer at the start of each week. Then I would set clear rules for creating labels and use only inputs available at prediction time.
- Tie the target to the decision the prediction will support.
- Define the entity, prediction time, and time window each row represents.
- Specify how outcomes become labels, including cases where outcomes are not yet known.
- Keep future information out of inputs to prevent data leakage.
CodeGoal: identify customers who may cancel a subscription. Target: whether an active customer cancels within the next 30 days. Granularity: one row per active customer each Monday. Inputs: information available before that Monday's prediction. Labels: assigned after the full 30-day outcome window.What the interviewer checks: The interviewer is checking whether you can turn a business goal into a clear target and consistent rows without data leakage.
Other ways it was asked
- “How would you define the exact prediction target and data granularity for this system?”
- “How would you define the prediction target and data granularity for a predictive system?”
What is the difference between OLTP and OLAP systems?
Asked in 4 of 274 interviews · last asked Sep 18 · Concept · Junior
Show answer
OLTP systems handle daily transactions, such as orders and payments. They support many short reads and writes, often using normalized tables. OLAP systems support reports and analysis with large scans and summaries, often using star schemas. OLTP focuses on fast, reliable transactions; OLAP focuses on efficient analysis.
- OLTP means online transaction processing; OLAP means online analytical processing.
- OLTP usually works with current operational data; OLAP often combines historical data from several sources.
- OLTP queries usually touch few rows; OLAP queries often aggregate many rows.
- Normalized tables and star schemas are common choices, not strict requirements.
What the interviewer checks: The interviewer checks whether you can link each workload to its query patterns and data model.
Other ways it was asked
- “What is the difference between OLAP and OLTP?”
- “What are OLAP and OLTP systems, and how do they differ?”
- “What's the difference between OLAP and OLTP?”
What is medallion architecture, and what are the bronze, silver, and gold layers?
Asked in 4 of 274 interviews · last asked Oct 24 · Concept · Mid-level
Show answer
Medallion architecture organizes data into layers with increasing quality and readiness for use. Bronze stores raw data as received from sources. Silver cleans, validates, and combines data into reliable records. Gold shapes data for business needs, such as reports, dashboards, and analysis.
- Bronze keeps source data so it can be traced and processed again.
- Silver handles duplicates, data types, missing values, and other quality issues.
- Gold often contains business metrics, summary tables, or dimensional models.
- The layers describe data quality and purpose, not a required database technology.
What the interviewer checks: The interviewer is checking whether you understand how raw data becomes trusted, business-ready data.
Other ways it was asked
- “What is medallion architecture, and how does it work?”
- “What is a medallion architecture, and how do its layers organize data?”
- “What is the medallion architecture for data processing?”
What is Apache Iceberg, and what are its key features for managing data lake tables?
Asked in 4 of 274 interviews · last asked Jul 16 · Concept · Senior
Show answer
Apache Iceberg is an open table format for data lakes, not a database engine. It tracks data files through metadata and snapshots, supporting ACID transactions and consistent reads. Its key features include schema and partition evolution, time travel, and efficient queries across large tables.
- Atomic snapshot updates support concurrent readers and writers without exposing partial changes.
- Schema evolution supports changes such as adding or renaming columns without rewriting existing data.
- Hidden partitioning simplifies queries; partition layouts can change without rewriting old files.
- Engines such as Spark, Trino, and Flink can share tables, though feature support varies.
What the interviewer checks: The interviewer is checking whether you understand how Iceberg adds reliable table management to data lake files.
Other ways it was asked
- “What have you built with Apache Iceberg, and how would you get up to speed using it as the table format across a data lake?”
- “What problems does Apache Iceberg solve that Delta Lake does not?”
- “How is Apache Iceberg different from Hive external tables?”
What is Delta Lake, and how do Delta tables support ACID transactions, schema enforcement, and time travel?
Asked in 4 of 274 interviews · last asked Oct 1 · Concept · Junior
Show answer
Delta Lake is an open-source storage layer that adds reliability to data lakes. Delta tables use Parquet files and a transaction log to provide ACID transactions, so writes commit fully or not at all. Schema enforcement rejects incompatible writes unless supported schema changes are explicitly allowed. Time travel reads earlier table versions while the required history and files remain available.
- ACID means atomicity, consistency, isolation, and durability; the transaction log tracks committed table versions.
- Readers see consistent snapshots, and conflicting writes may fail rather than corrupt data.
- Schema enforcement checks write compatibility; schema evolution allows supported schema changes.
- Time travel supports versions or timestamps; VACUUM can remove files needed to read older versions.
What the interviewer checks: The interviewer wants to know whether you understand how Delta adds reliable writes, schema control, and historical reads.
Other ways it was asked
- “What is Delta Lake, and how does it support ACID transactions, schema enforcement, and time travel?”
- “Can you tell me what advantages Delta Tables bring?”
- “What is Delta Lake in Databricks?”
When would you denormalize a database, and what trade-offs would you consider?
Asked in 3 of 274 interviews · last asked Sep 5 · Concept · Mid-level
Show answer
I would denormalize when measured read performance problems justify storing duplicate or precomputed data, such as totals for frequent reports. I would first check whether indexes or query changes solve the problem. The trade-offs are extra storage, more work on writes, and the risk of stale or inconsistent data.
- Useful for read-heavy reports, dashboards, and queries with costly joins.
- Faster reads can come at the cost of slower, more complex writes.
- Define how duplicate data stays in sync and how much delay is acceptable.
- Measure the benefit against the added storage and maintenance work.
What the interviewer checks: The interviewer is checking whether you can justify denormalization with workload needs and manage its consistency costs.
Other ways it was asked
- “When would you intentionally denormalize a data model?”
- “In what real world scenarios would you intentionally denormalize a database?”
- “What are some caveats of denormalizing tables?”
How would you design a dimensional model to solve a business problem, choose the fact table grain, and evaluate the trade-offs?
Asked in 3 of 274 interviews · last asked Sep 29 · Design · Senior
Show answer
To compare sales and profit by product, store, and day, I would build a sales star schema. Each fact row would represent one sale line, with quantity, revenue, cost, and dimension keys. Dimensions would hold descriptive details, such as product category. This grain supports detailed analysis but needs more storage and query work than daily totals.
- Choose the grain from business questions and available source detail; define exactly what one row represents.
- Keep every measure consistent with the grain; do not repeat order totals on each sale line.
- Separate numeric measures from descriptive dimensions so reports can group and filter facts clearly.
- Decide whether reports need historical dimension values; preserving changes adds complexity.
What the interviewer checks: The interviewer checks whether you can connect business needs to a clear grain, correct measures, and practical trade-offs.
Other ways it was asked
- “Tell me about a dimensional model you designed. How did you choose the fact table grain, and why did you separate facts from dimensions?”
- “Tell me about a dimensional model you designed. What business problem did it solve, how did you choose the fact table grain, and what trade-offs did you consider?”
- “Tell me about the dimensional model you designed, and how did you decide the fact table grain and what trade-offs did that decision create?”
What's the difference between a fact table and a dimension table?
Asked in 3 of 274 interviews · last asked Sep 18 · Concept · Junior
Show answer
A fact table stores business events or measurements, such as sales quantities and amounts. A dimension table stores details that describe those facts, such as product names and categories. Fact tables link to dimension tables through keys, so we can filter and group the facts.
- Define the fact table’s grain: exactly what one row represents.
- Fact tables usually contain numeric measures and foreign keys to dimensions.
- Dimensions provide descriptive fields for filtering, grouping, and labeling reports.
- Some fact tables record events without numeric measures; these are called factless fact tables.
CodeSales fact: sale_line_id, product_id, quantity, sales_amount Product dimension: product_id, product_name, category Each Sales row represents one sale line and links to one Product row.What the interviewer checks: The interviewer checks whether you can separate measurements from descriptive details and explain how they connect.
Other ways it was asked
- “What is the difference between a fact table and a dimension table?”
- “What are dimension tables and fact tables, and how do they differ?”
How would you handle corrections and repeated updates without creating duplicate fact records?
Asked in 3 of 274 interviews · last asked Sep 21 · Scenario · Mid-level
Show answer
I define what one fact row represents and give it a stable business key. I use that key to update an existing row or insert a new one, with a unique constraint to prevent duplicates. I track the source version so repeated or older updates cannot overwrite newer data. If history is needed, I store each version separately.
- The key must identify one fact, such as a shipment event, not just the shipment.
- Deduplicate incoming updates before applying them.
- Reprocessing the same update should leave the result unchanged.
- Keep correction history separate from the current fact, or clearly mark the current version.
What the interviewer checks: The interviewer checks whether you can define fact identity and safely handle retries, corrections, and out-of-order updates.
Other ways it was asked
- “How would you prevent multiple updates to the same shipment from creating duplicate fact records?”
- “How would you model corrections to an existing operational event and prevent duplicate fact records?”
- “How would you handle corrections or repeated updates without creating duplicate fact records?”
What is a data warehouse?
Asked in 3 of 274 interviews · last asked Oct 19 · Concept · Junior
Show answer
A data warehouse is a system that stores data for reporting and analysis. It brings together cleaned data from different sources and keeps historical records. It is designed for analytical queries rather than day-to-day tasks such as placing orders.
- Combines data from multiple sources using consistent definitions and formats.
- Keeps historical data to support trend analysis.
- Supports reporting and complex queries, rather than routine transaction processing.
What the interviewer checks: The interviewer is checking whether you understand a data warehouse’s purpose and how it differs from a transaction database.
How would you design Salesforce data storage to handle millions of records?
Asked in 3 of 274 interviews · last asked Oct 1 · Design · Senior
Show answer
I would keep active Accounts, Contacts, and Cases in standard objects, with relationships based on business needs. I would plan for yearly growth, storage limits, access rules, and retention, then archive older records or store them outside Salesforce. I would test queries and bulk operations at the expected scale.
- Use selective filters on indexed fields, and check query plans with realistic data.
- Avoid concentrating too many records under one parent or owner; this can cause locking and sharing problems.
- Consider Big Objects or external storage, but check their query, reporting, and access limits first.
- For migration, use Bulk API and external IDs for upserts; load parents first and validate results.
What the interviewer checks: The interviewer checks whether you can balance Salesforce storage, performance, access, and data lifecycle needs at large scale.
Other ways it was asked
- “What should you consider when designing a Salesforce data model for millions of Account, Contact, and Case records?”
- “What should you consider when designing a Salesforce data model for millions of Account, Contact, and Case records created annually?”
- “How would you migrate millions of records from a legacy CRM to Salesforce, and how would you model and store very large data volumes in Salesforce?”
When modeling a new table in PostgreSQL, how do you decide between a natural key and a surrogate key, such as BIGSERIAL or UUID, and what performance or operational trade-offs drive that choice?
Asked in 2 of 274 interviews · last asked Jul 27 · Concept · Senior
Show answer
I use a natural key when it is small, stable, and reliably unique. Otherwise, I use a surrogate key and enforce business uniqueness separately. BIGINT identity or BIGSERIAL gives compact indexes and mostly sequential inserts. UUIDs allow independent ID generation across systems, but use more space and, when random, have poorer index locality.
- Changing a natural primary key can require updates to foreign keys in other tables.
- UUID uses 16 bytes; BIGINT uses 8, affecting indexes and foreign key columns.
- Identity and BIGSERIAL use sequences: gaps are normal, and imports may require resetting the sequence.
- Time-ordered UUIDs improve index locality compared with random UUIDs, but remain larger than BIGINT.
What the interviewer checks: The interviewer checks whether you balance business uniqueness, key stability, database performance, and ID generation across systems.
Other ways it was asked
- “When you are designing tables in PostgreSQL, what factors make you choose a UUID primary key versus a big serial or identity key?”
What relationship types are available in Dynamics 365?
Asked in 2 of 274 interviews · last asked Apr 16 · Concept · Junior
Show answer
In Dynamics 365 apps that use Dataverse, entity relationships are one-to-many (1:N), many-to-one (N:1), and many-to-many (N:N). Many-to-one is a one-to-many relationship viewed from the other side. One-to-many uses a lookup column, while native many-to-many uses an intersect table.
- One-to-many: one account can have many contacts; the lookup is on the contact.
- Many-to-one: many contacts can link to the same account.
- Many-to-many: records on both sides can have multiple links, such as users and security roles.
What the interviewer checks: The interviewer checks whether you understand relationship direction and how Dataverse links records.
Other ways it was asked
- “What types of entity relationships are available in Microsoft Dynamics 365?”
What are data models typically used for?
Asked in 2 of 274 interviews · last asked Oct 6 · Concept · Junior
Show answer
Data models describe how data is organized, how items relate, and what rules the data must follow. They help teams agree on business needs and design databases for storing, querying, and reporting data. They also guide changes as those needs grow.
- Use a conceptual model to agree on main business concepts and relationships.
- Use a logical model to define detailed entities, attributes, keys, and relationships without choosing a specific database.
- Use a physical model to define storage details, such as tables, columns, and indexes, for a chosen database.
What the interviewer checks: The interviewer checks whether you can connect data models to business needs and explain their main levels.
Other ways it was asked
- “What types of data models are commonly used, and when would you choose each?”
- “What are data models?”
What is an ER diagram?
Asked in 2 of 274 interviews · last asked Jul 19 · Concept · Junior
Show answer
An ER diagram, or entity-relationship diagram, shows the entities in a data model, their attributes, and how they relate. It also shows how many instances of one entity can relate to another. It helps explain and plan a database structure before building it.
- Entities represent things such as customers or orders; attributes describe them.
- Relationships show how entities connect.
- Cardinality shows one-to-one, one-to-many, or many-to-many relationships.
- Optionality shows whether a relationship is required or optional.
CodeCustomer (customer_id, name) Order (order_id, order_date, customer_id) One customer can have zero or many orders. Each order belongs to exactly one customer.What the interviewer checks: The interviewer is checking whether you understand the main parts of an ER diagram and its role in database design.
Data modeling basics interview questions
Review entities, ER diagrams, relationships, keys, constraints, normalization, denormalization, and basic metadata. 52 questions.
What is a dataset?
Asked in 2 of 274 interviews · last asked Oct 4 · Concept · Junior
Other ways it was asked
- “What is a data set?”
What is the business purpose of turning customer activity into structured database fields?
Asked once · last asked Jul 22 · Concept · Mid-level
Other ways it was asked
- “What is the purpose of compressing customer activity into structured database fields?”
- “What is the purpose of compressing customer information into structured database data?”
- “What is the business purpose of turning customer activity into structured data?”
How would you apply data modeling techniques in an analytics project, and how would you assess their impact on the outcome?
Asked once · last asked Feb 13 · Behavioral · Senior
Other ways it was asked
- “How did you use data modeling techniques in a recent analytics project, and what impact did they have on the project outcome?”
In a one-to-many relationship between profiles and phone numbers, which table should contain the foreign key?
Asked once · last asked Feb 23 · Concept · Junior
How would you prevent the same phone number from being assigned to multiple user profiles?
Asked once · last asked Feb 23 · Scenario · Mid-level
What is the difference between structured and unstructured data?
Asked once · last asked Mar 28 · Concept · Junior
When creating a one-to-many relationship, what relationship behaviors can you specify?
Asked once · last asked Apr 4 · Concept · Mid-level
If an entity has a unique attribute, such as a passport number, should you still create a separate primary key?
Asked once · last asked Apr 27 · Concept · Mid-level
How can you implement a one-to-one relationship between two database tables?
Asked once · last asked Apr 27 · Concept · Mid-level
What is the difference between identifying and non-identifying relationships in data modeling?
Asked once · last asked May 16 · Concept · Junior
How do you apply normalization when designing a relational data model?
Asked once · last asked May 16 · Concept · Junior
Is storing multiple related IDs in an array really the best way to model a many-to-many relationship?
Asked once · last asked Jul 17 · Concept · Mid-level
Other ways it was asked
- “What is a better database design for representing many-to-many relationships than storing related IDs in an array?”
- “What alternative to storing multiple related IDs in an array would you use to model these relationships?”
Database schema design interview questions
Work through relational, document, and graph model designs for applications, business workflows, industry domains, and machine learning datasets. 141 questions.
How would you design users, roles, and matching at the database level?
Asked in 2 of 274 interviews · last asked Jul 22 · Design · Mid-level
What are the differences between the code-first and database-first approaches in Entity Framework, and when would you use each?
Asked in 2 of 274 interviews · last asked Oct 9 · Concept · Mid-level
Other ways it was asked
- “What are the differences between code-first and database-first approaches in Entity Framework Core?”
How would you design and implement a wish-list feature for an e-commerce application, considering data storage, the user interface, and scalability?
Asked once · last asked Mar 31 · Design · Mid-level
Other ways it was asked
- “What challenges might arise when developing a wish-list feature for an e-commerce application, and how would you address them?”
What information would you store from past tickets in the memory of a ticket-handling agent?
Asked once · last asked Aug 7 · Design · Senior
Other ways it was asked
- “What metadata would you store for past tickets in an agent's memory?”
Can you name some basic tables in the TriZetto Facets claims data model?
Asked once · last asked Aug 26 · Concept · Junior
Other ways it was asked
- “Can you name some basic tables in the TriZetto Facets membership data model?”
How can MongoDB be used in a doctor appointment booking system?
Asked once · last asked Sep 10 · Design · Mid-level
How would you define a unique operational event in a data model?
Asked once · last asked Sep 21 · Concept · Mid-level
How would you model and implement relationships among projects, team members, and tasks in PostgreSQL?
Asked once · last asked Sep 30 · Concept · Mid-level
Other ways it was asked
- “How would you model the relationships among projects, team members, and tasks in a relational database?”
- “What is the relationship between projects and team members in a relational database?”
- “What is the relationship between tasks and projects in a relational database?”
Should a source identifier be stored on each calculation event or on a higher-level object?
Asked once · last asked Oct 9 · Concept · Mid-level
Other ways it was asked
- “Should a source identifier be stored on each calculation event or on a higher-level object to avoid repeating it across object blocks?”
How would you design a stock-on-hand (SOH) and available-to-promise (ATP) status model and audit logs for inventory reservations?
Asked once · last asked Oct 17 · Design · Senior
Other ways it was asked
- “How would you design a stock-on-hand (SOH) and available-to-promise (ATP) status model and audit logs for reservations, picks, packing, shipping, cancellations, and returns across POS, OMS, and WMS?”
What is the best practice when a system contains both dynamic data and data that does not change?
Asked once · last asked Feb 5 · Concept · Mid-level
How would you create custom CRM objects and associate them with other objects, such as linking an event object to companies?
Asked once · last asked Feb 20 · Tooling · Mid-level
Dimensional modeling interview questions
These questions cover business requirements, fact and dimension tables, grain, star and snowflake schemas, hierarchies, and measures. 28 questions.
How do you choose between event-level data, daily summaries, and current-state models?
Asked in 2 of 274 interviews · last asked Sep 21 · Concept · Mid-level
Other ways it was asked
- “Why would you choose an event-level grain instead of a daily aggregated fact table?”
- “How do you decide whether to model data at the event level, as a daily summary, or as a current-state model, and what makes that decision difficult?”
In a star schema, when do you choose a surrogate key over a natural key for a dimension?
Asked once · last asked Aug 23 · Concept · Mid-level
Other ways it was asked
- “In a star schema, when would you choose a surrogate key over a natural key for a dimension, and how would you maintain that choice during incremental loads?”
How do you model data in a warehouse using fact and dimension tables, and how do you design schemas?
Asked once · last asked Apr 14 · Concept · Mid-level
Where would you use a star schema? Can you give an example of an application where it might be used?
Asked once · last asked Jun 11 · Concept · Junior
How do business needs and user personas influence the choice of data model?
Asked once · last asked Jul 16 · Concept · Mid-level
How do you design dbt gold-layer models, including dimensions, facts, and dataset selection?
Asked once · last asked Jul 28 · Scenario · Senior
How would you maintain relationships in a length-of-stay data mart?
Asked once · last asked Sep 19 · Design · Mid-level
How do you load dimension and fact tables in Snowflake using a star schema?
Asked once · last asked Sep 19 · Tooling · Mid-level
How do you translate a grain definition into a data model?
Asked once · last asked Sep 20 · Concept · Mid-level
What are dimensions and measures in data analysis, and how are they used?
Asked once · last asked Sep 21 · Concept · Junior
What trade-offs can arise when choosing a fact table grain and separating facts from dimensions in a dimensional model?
Asked once · last asked Sep 21 · Concept · Mid-level
How would you model revenue by both Order Date and Ship Date using one Date dimension?
Asked once · last asked Sep 22 · Scenario · Mid-level
Show 13 more
If you were analyzing farming interactions over time by region, what grain would you choose for the central fact table, and how would you handle a slowly changing dimension?
Asked once · last asked Sep 23 · Design · Senior
If you need both shipment-event reporting and a daily view of warehouse capacity, would you put those measures in one fact table? Walk me through the fact table or tables you would choose and how you would prevent double counting.
Asked once · last asked Sep 28 · Design · Senior
If you had to define supply-risk dataset requirements with data engineering, what specific artifact would you give them, and how would you check during development that the dataset still answers the business question?
Asked once · last asked Sep 29 · Scenario · Senior
Operations needs one row per shipment to investigate delays, while finance needs one row per customer per month for a summary report. What grain would you use for the core fact table, and what would you give engineering so both teams get correct numbers?
Asked once · last asked Sep 29 · Design · Senior
If the request is only to show which suppliers are risky, what would you clarify before modeling or creating a chart, and what would your first production-ready version contain?
Asked once · last asked Sep 29 · Scenario · Senior
Tell me about a project where you translated a business need into a reliable dataset and report. What design decisions did you make, and how did you validate the result?
Asked once · last asked Oct 1 · Behavioral · Mid-level
If a $1,000 order is split into three shipments and only two are dispatched this month, how would you clarify the business requirement and determine what data is needed to calculate the value shipped this month accurately?
Asked once · last asked Oct 1 · Scenario · Mid-level
How would you model daily sales for about 1,400 stores, more than 20 retail chains, and multiple SKUs for analysis?
Asked once · last asked Oct 2 · Design · Mid-level
How would you model a report that combines sales by store and day with promoter hours by promoter, store, supplier, and day?
Asked once · last asked Oct 2 · Design · Senior
How do you maintain fact and dimension tables?
Asked once · last asked Sep 19 · Scenario · Mid-level
Other ways it was asked
- “Can you give an example of how you maintain fact and dimension tables?”
What is data granularity?
Asked once · last asked Mar 28 · Concept · Junior
What is a star schema?
Asked once · last asked Jun 11 · Concept · Junior
How would you write a SQL CTE using sales and product tables in a star schema?
Asked once · last asked Sep 14 · Write code · Mid-level
Slowly changing dimensions and historical data interview questions
Explore how to model changing data with SCD types, effective dates, snapshots, audit trails, corrections, and historically accurate joins. 34 questions.
What is the difference between SCD Type 1 and SCD Type 2?
Asked in 2 of 274 interviews · last asked Sep 20 · Concept · Mid-level
Other ways it was asked
- “What are SCD Type 1 and Type 2, and how do they differ?”
When is using yesterday’s customer dimension safe for today’s transaction feed, and what checks and approvals are required?
Asked once · last asked Sep 29 · Scenario · Senior
Other ways it was asked
- “If SQL checks find no transactions linked to today’s duplicate IDs, but today’s feed includes 200 new customers who were not in yesterday’s dimension, would you use yesterday’s dimension, and what would you tell the service owner?”
- “If today's fact contains transactions for 200 customers absent from yesterday's dimension and the duplicate keys have no matching transactions, would you use yesterday's dimension for the 5 p.m. feed, or what would you tell the service owner?”
- “Under what exact conditions would you recommend using yesterday’s customer dimension for today’s transaction feed, and who would approve that decision?”
You need to store each customer's product price history for 1–20 years. Which type of database would you choose, and why?
Asked once · last asked Apr 15 · Scenario · Mid-level
How would you map customers_src (customer ID, name, country) to warehouse table customers_dim (customer key, customer ID, name, country, is_active), including logic for populating customer keys and is_active and handling new versus existing customers?
Asked once · last asked May 18 · Scenario · Mid-level
Other ways it was asked
- “Suppose you have a source table called customers_src and a target table called customers_dim in a data warehouse. The target requires customer key, customer ID, name, country, and is_active, but the source provides only customer ID, name, and country. How would you design a simple mapping and logic to populate the target table, including handling new versus existing customers?”
How would you apply an update to custom classification logic to events already stored in a normalized database?
Asked once · last asked Jul 20 · Design · Senior
How does SCD Type 2 represent changes in a dataset?
Asked once · last asked Aug 18 · Concept · Junior
Can you provide a four- or five-row example dataset showing what SCD Type 2 records look like?
Asked once · last asked Aug 18 · Concept · Junior
Can you show how a target table is populated during an incremental load for employee ID and employee name, including the date columns?
Asked once · last asked Aug 18 · Scenario · Mid-level
In an SCD Type 2 pipeline, how would you set the start date, end date, and active flag using the current date?
Asked once · last asked Aug 18 · Concept · Mid-level
How would you separate actual and historical data in an SAP Datasphere solution?
Asked once · last asked Aug 28 · Scenario · Mid-level
When a BigQuery MERGE statement updates existing records and inserts new ones, does that implement SCD Type 1 or Type 2?
Asked once · last asked Sep 10 · Concept · Mid-level
What are snapshots in dbt?
Asked once · last asked Sep 14 · Concept · Mid-level
Show 20 more
How would you handle an audit log requirement when the same order gets updated over time?
Asked once · last asked Sep 18 · Scenario · Mid-level
How would you model several updates for the same operational event when its status or timestamp changes more than once?
Asked once · last asked Sep 21 · Scenario · Mid-level
How would you handle repeated deliveries or later corrections in a dimensional model without changing the fact-table grain or creating duplicate fact records?
Asked once · last asked Sep 21 · Scenario · Senior
How would you define a unique operational event and prevent repeated deliveries or later corrections from creating duplicate fact records?
Asked once · last asked Sep 22 · Scenario · Senior
How would you distinguish a repeated delivery of the same event from a new business event or a correction to an earlier event?
Asked once · last asked Sep 22 · Scenario · Senior
If a farm moves from one region to another in July and a manager asks for last year’s interactions grouped by the farm’s current region, how would you model that while avoiding confusion with historical region reporting?
Asked once · last asked Sep 23 · Scenario · Senior
A shipment arrives late Monday, but its WMS record arrives just after midnight Tuesday; which timestamp would you use to match it to a daily capacity snapshot, and how would you handle snapshots that record only end-of-day capacity?
Asked once · last asked Sep 28 · Scenario · Senior
Other ways it was asked
- “A shipment arrives late on Monday, but its WMS record is received just after midnight on Tuesday. Which timestamp would you use to match the shipment to the daily capacity snapshot, and what would you do if the snapshot only records capacity at the end of each day?”
A customer's risk changes from medium to high; how would you model a Monday transaction loaded Friday to reflect risk at transaction time, and what would you publish if the source provides only current risk, without history or effective dates?
Asked once · last asked Sep 28 · Scenario · Senior
Other ways it was asked
- “Suppose a customer’s risk category changes from medium to high, but a transaction occurred on Monday and was loaded late on Friday. How would you model the customer dimension so the Monday transaction reflects the category that applied when it occurred? If the source only gives the current category, with no effective date or history, can you assert that the category was medium on Monday, and what would you publish?”
If the unit conversion changed from 12 to 10 units last month, how would you preserve historically correct quantities when rerunning and reporting on past periods?
Asked once · last asked Sep 28 · Scenario · Senior
How would you prevent repeated ETA changes from making a supplier look reliable?
Asked once · last asked Sep 29 · Scenario · Senior
How would you report historical customer risk when the source provides only the current classification?
Asked once · last asked Sep 29 · Scenario · Senior
Other ways it was asked
- “If a customer-risk history report is required but the source only supplies each customer’s current risk category and has no change log, can you report a Monday transaction as high risk when the customer is marked high risk later? What would you put in the model and tell the business analyst?”
- “If the only available customer risk category is the current value, marked high on Friday, and a Tuesday shipment is due in a historical risk report, what value would you put in the risk-category field, and how would you explain it to the service owner?”
If a shipment was received Tuesday but ingested Friday, and two customer-dimension rows have the same business key with different validity periods, which dimension row should you join to, and how would you test that choice?
Asked once · last asked Sep 29 · Scenario · Senior
If both mappings point to the same region but their effective date ranges overlap, would you keep the latest mapping, and why or why not?
Asked once · last asked Oct 1 · Scenario · Senior
How do you handle slowly changing dimensions (SCD types) in dbt?
Asked once · last asked Oct 2 · Tooling · Mid-level
A product can change suppliers over time, and a store can change banners. Should a 2025 report show historical or current values, and how would you implement each option?
Asked once · last asked Oct 2 · Scenario · Senior
How would you implement supplier history over time in a data model?
Asked once · last asked Oct 2 · Scenario · Senior
When would you use slowly changing dimensions (SCD) to preserve supplier or store-banner history?
Asked once · last asked Oct 2 · Concept · Mid-level
How would you change a courier's delivery-platform assignment in a system while preserving their historical data?
Asked once · last asked Oct 7 · Scenario · Mid-level
How should a functional specification define courier status transitions so that historical data is preserved?
Asked once · last asked Oct 7 · Scenario · Mid-level
How would you track an order as it moves through a sequence of states?
Asked once · last asked Oct 15 · Concept · Mid-level
Data warehouse architecture interview questions
Compare warehouses, lakes, and marts, then review OLTP, OLAP, medallion layers, ETL, CDC, cloud platforms, and migration planning. 78 questions.
How do a record’s schema and content change through the bronze, silver, and gold layers?
Asked in 2 of 274 interviews · last asked Sep 17 · Concept · Mid-level
Other ways it was asked
- “What changes to a record as it moves through the three layers of the medallion architecture?”
- “What are the concrete differences in schema and content between the silver and gold layers of a table in a medallion architecture?”
- “What preprocessing steps are typically performed in the silver layer of a medallion architecture?”
What is your product vision for a data platform, specifically around data management?
Asked once · last asked Sep 7 · Senior
Other ways it was asked
- “What would be your product vision for a data platform, including data management and datasets?”
What is your approach to providing an end-to-end data warehouse solution for a client?
Asked once · last asked Sep 19 · Design · Mid-level
Other ways it was asked
- “What would be your approach to implementing a data warehouse from scratch and delivering an end-to-end solution?”
How would you create a healthcare data warehouse?
Asked once · last asked Sep 19 · Design · Mid-level
Other ways it was asked
- “How would you create a healthcare data warehouse for a client?”
- “How do you create a data warehouse?”
What is a data mart, and why can it make queries faster?
Asked once · last asked Sep 19 · Concept · Junior
Other ways it was asked
- “What is a data mart, and why is it faster?”
- “Why are data marts faster?”
What criteria would you use to choose Snowflake over other data warehouse solutions?
Asked once · last asked Sep 20 · Scenario · Mid-level
What source types and data models, such as fact, dimension, and event tables, are commonly used in data pipelines?
Asked once · last asked Mar 3 · Concept · Mid-level
What data-loading patterns do you know for loading raw data into a warehouse?
Asked once · last asked Mar 20 · Concept · Mid-level
What is columnar storage, and why is it commonly used in data warehouses?
Asked once · last asked Apr 14 · Concept · Mid-level
What is the Data Vault modeling approach, and when would you use it?
Asked once · last asked Apr 17 · Concept · Senior
How would you design a CDC pipeline using medallion architecture for nightly data from different transactional databases, including processing and best practices at the bronze, silver, and gold layers?
Asked once · last asked Apr 24 · Design · Senior
Other ways it was asked
- “How would you design a CDC data pipeline using medallion architecture for nightly data from different transactional databases, and what would you do at each layer—from the raw bronze layer through silver to gold? What best practices would you follow?”
What types of operations are performed in online analytical mining (OLAM)?
Asked once · last asked May 2 · Concept · Mid-level
Lakehouse, Apache Iceberg, and Delta Lake interview questions
Lakehouse questions cover Parquet, Avro, Iceberg, Delta Lake, catalogs, storage, time travel, maintenance, and access across engines. 43 questions.
How does Apache Iceberg store data internally?
Asked in 2 of 274 interviews · last asked Aug 20 · Concept · Senior
Other ways it was asked
- “How does Apache Iceberg work, specifically regarding data storage and search functionality?”
What is the difference between external tables and managed tables?
Asked in 2 of 274 interviews · last asked Aug 20 · Concept · Junior
Other ways it was asked
- “What is the difference between managed and external tables in Databricks?”
- “What are the benefits of managed tables versus external tables?”
When should you run VACUUM on a Delta table, and what does it remove?
Asked once · last asked Apr 17 · Tooling · Mid-level
Other ways it was asked
- “What does the VACUUM command do to a Delta table?”
What types of Apache Iceberg catalog integrations are available, and where can the catalog be stored?
Asked once · last asked Jun 9 · Concept · Senior
Other ways it was asked
- “Where can an Apache Iceberg catalog be hosted?”
How can Apache Iceberg data be exposed through a REST API?
Asked once · last asked Jun 9 · Design · Senior
What is the use case for Iceberg, and why should I use it?
Asked once · last asked Jun 9 · Concept · Mid-level
What do you understand by a lakehouse, how does Iceberg fit into it, and can you design a lakehouse without Iceberg or with something else?
Asked once · last asked Jun 9 · Design · Senior
If you built an Iceberg lakehouse, how would you expose its data to downstream consumers outside AWS so they could access, download, and work with it?
Asked once · last asked Jun 9 · Design · Senior
How would consumers connect to a file-based Apache Iceberg lakehouse using JDBC?
Asked once · last asked Jun 9 · Design · Senior
Other ways it was asked
- “Where is the metadata required for JDBC access to a file-based Apache Iceberg lakehouse hosted?”
- “Where should a JDBC query be sent when querying an Apache Iceberg lakehouse?”
What solution would you use to expose Apache Iceberg data to someone outside AWS who does not have EMR?
Asked once · last asked Jun 9 · Design · Senior
How does schema evolution work in Apache Iceberg?
Asked once · last asked Jun 9 · Concept · Senior
If an Apache Iceberg manifest file were deleted, what would happen to the data, and would you still be able to access it?
Asked once · last asked Jun 9 · Scenario · Senior
Other ways it was asked
- “Can deleting an Apache Iceberg manifest file cause data corruption?”
Semantic modeling and Power BI interview questions
Check your understanding of shared reporting models, relationships, calculations, and aggregation in Power BI, Microsoft Fabric, and SAP. 52 questions.
For a one-to-many relationship between two tables, which SQL join would you use to get the same results as Power BI?
Asked once · last asked Aug 28 · Concept · Mid-level
What can happen when a Power BI data model has a many-to-many relationship?
Asked once · last asked Sep 2 · Concept · Mid-level
Other ways it was asked
- “What can happen when both tables in a Power BI data model have a many-to-many relationship?”
What are dimensions and measures in Power BI, and how are they used?
Asked once · last asked Sep 21 · Concept · Junior
Other ways it was asked
- “What is the difference between dimensions and measures in Power BI?”
How do you validate and enforce consistent metric definitions in Power BI—for example, revenue, AOV, and active customer—across multiple reports and teams?
Asked once · last asked Jun 25 · Tooling · Senior
When a dashboard needs to compare multiple datasets with different schemas, how do you standardize metrics and visuals in Excel so trends remain comparable without confusing users?
Asked once · last asked Jul 29 · Tooling · Senior
How would you use Power Query and Power Pivot to join data and calculate monthly segment margins, and what choices would you make to optimize refresh performance?
Asked once · last asked Aug 1 · Tooling · Senior
Other ways it was asked
- “How would you use Power Query and Power Pivot to do that join and monthly segment margin, and what refresh performance choices would you make?”
What is a composite model in Power BI, and when would you use one?
Asked once · last asked Sep 4 · Concept · Mid-level
What factors affect semantic model performance?
Asked once · last asked Sep 4 · Concept · Senior
What is the best way to build multiple reports in Microsoft Fabric on a shared semantic model?
Asked once · last asked Sep 9 · Design · Mid-level
What challenges should you consider when migrating reports from Tableau to Power BI?
Asked once · last asked Sep 9 · Scenario · Mid-level
What limitations do you see in Power BI compared with Tableau for data modeling?
Asked once · last asked Sep 9 · Concept · Mid-level
In Microsoft Fabric, can you use a snowflake schema in Power BI while maintaining report performance, and how do modeling limitations affect that choice?
Asked once · last asked Sep 9 · Scenario · Senior
Database performance and scalability interview questions
Performance questions focus on partitioning, clustering, indexes, query plans, sharding, tenant isolation, large Salesforce datasets, and reporting workloads. 35 questions.
Can you tell me about partitioning and how you select the right column for partitioning a table?
Asked in 2 of 274 interviews · last asked Sep 10 · Concept · Mid-level
Other ways it was asked
- “How would you choose columns for partitioning and clustering in a table?”
How would you choose between shared tables and separate databases or tables for a multi-tenant application?
Asked in 2 of 274 interviews · last asked Oct 2 · Design · Senior
Other ways it was asked
- “How would you manage tenant data in a multi-tenant application?”
- “How would you manage the database for a multi-tenant application?”
- “Which database would you choose for a new multi-tenant project, and how would you implement the multi-tenancy?”
In a query joining a large fact table to multiple dimensions, what indicators tell you the plan is inefficient?
Asked once · last asked Aug 23 · Scenario · Senior
Other ways it was asked
- “When a query joining a large fact table to multiple dimensions has an inefficient execution plan, what indicators would you look for, and what indexing or partitioning changes would you try first?”
What multitenancy patterns or data-segmentation approaches can you use besides assigning a tenant ID?
Asked once · last asked Feb 10 · Concept · Senior
How would you design passport-number lookups and candidate score updates to avoid searching every company?
Asked once · last asked Mar 6 · Design · Senior
When building a secondary index for candidates by passport number, what should each index entry store?
Asked once · last asked Mar 6 · Concept · Mid-level
For a read-only REST service that retrieves records containing a BLOB and fails under 30 to 100 simultaneous requests, what changes would you consider to the database table structure?
Asked once · last asked Mar 17 · Scenario · Senior
How would you approach database schema design and performance tuning?
Asked once · last asked Apr 9 · Concept · Mid-level
Other ways it was asked
- “Can you tell me more about your schema design and any performance tuning you've done with your databases?”
How would you design a rough architecture for an image management system used by more than 50,000 merchants with over 4 billion images? What data would you need to manage, and how would you manage it?
Asked once · last asked Apr 13 · Design · Senior
When might hash-based sharding be unsuitable, even if it distributes data evenly across shards?
Asked once · last asked Apr 13 · Scenario · Senior
Explain your strategy for partitioning in Azure Data Lake Storage to account for large-scale enterprise projects.
Asked once · last asked May 9 · Design · Senior
When you're designing endpoints and data models for a web application, how do you ensure that they're efficient and scalable for future growth?
Asked once · last asked Jul 16 · Design · Mid-level
Show 22 more
How do you handle very large JSON columns in a frequently read table? When do you keep JSON inline versus moving fields into typed columns or a side table, and why?
Asked once · last asked Jul 27 · Design · Senior
Would a 30-day time-to-live policy be appropriate for dropping old data partitions, and what factors would you consider?
Asked once · last asked Jul 29 · Scenario · Mid-level
How would you decide whether to partition a table, and how would you choose or change its partitioning strategy?
Asked once · last asked Jul 29 · Design · Mid-level
How would you scale data pipelines and enterprise data warehouse solutions?
Asked once · last asked Jul 29 · Design · Senior
What are the considerations for using indexes in Greenplum and other MPP systems?
Asked once · last asked Jul 31 · Concept · Senior
How would you structure data and use serverless SQL pools to ensure cost-effective query performance in Azure Synapse Analytics?
Asked once · last asked Aug 11 · Design · Senior
If most queries filter on order date, what would be your partitioning strategy for the orders table?
Asked once · last asked Aug 14 · Scenario · Mid-level
If both partitioning and liquid clustering are available, when would partitioning be the better choice?
Asked once · last asked Aug 14 · Scenario · Senior
If you need to delete historical data, would you prefer partitioning or liquid clustering, and why?
Asked once · last asked Aug 14 · Scenario · Senior
How would you structure tenant data if you did not want to store it all in one table?
Asked once · last asked Aug 20 · Design · Senior
When processing data in Databricks, what considerations would you account for with Delta Lake, partitioning, and performance optimization?
Asked once · last asked Aug 22 · Design · Senior
How would you prevent data science and BI workloads from interfering when they query the same underlying tables?
Asked once · last asked Aug 26 · Scenario · Senior
How would you handle tenant isolation so customer data doesn’t leak across tenants?
Asked once · last asked Sep 1 · Scenario · Senior
Can you explain why ClickHouse is good for analytics, and how it differs from Postgres?
Asked once · last asked Sep 2 · Concept · Mid-level
How would you choose an approach for working with a Power BI dataset containing millions of rows?
Asked once · last asked Sep 4 · Scenario · Senior
What problems can occur if a table is partitioned by country and one country has much more data than the others?
Asked once · last asked Sep 10 · Scenario · Senior
How can you fine-tune a Power BI dashboard with 10 million rows, a 3.5-hour refresh, and an 8 GB file size?
Asked once · last asked Sep 14 · Scenario · Senior
You are designing an org with 20,000,000 records and a few accounts will each have hundreds of thousands of related records. What risks would you assess and how will they influence your data model and sharing design?
Asked once · last asked Sep 18 · Design · Senior
Which metric would you track over a month to determine whether a data model optimization is holding up over time?
Asked once · last asked Sep 22 · Concept · Mid-level
If a database table is partitioned, when might splitting the data into separate tables still be necessary?
Asked once · last asked Sep 24 · Concept · Senior
How would you handle enterprise data architecture for millions of records across Salesforce and analytics needs?
Asked once · last asked Sep 30 · Design · Senior
How would you partition a very large dataset?
Asked once · last asked Jul 22 · Scenario · Mid-level
Data quality and governance interview questions
Consider shared definitions, source-of-truth decisions, reconciliation, duplicate keys, referential integrity, schema changes, data contracts, documentation, security, and privacy. 63 questions.
How would you define and enforce a consistent metadata schema across teams, collectors, and sessions?
Asked in 2 of 274 interviews · last asked Jul 25 · Scenario · Senior
Other ways it was asked
- “How do you design and enforce a Hindi metadata schema—including speaker demographics, device, environment, and consent—so it stays consistent across many collectors and sessions?”
- “How do you define and enforce a metadata schema, including required fields, naming rules, and controlled vocabulary across a team, so downstream analytics can reliably join sessions, builds, scenarios, and device configurations?”
How would you map platform-specific identifiers to a normalized campaign ID?
Asked once · last asked Jul 20 · Design · Mid-level
Other ways it was asked
- “How would you map platform-specific identifiers, such as Facebook campaign IDs or Google objective IDs, to a normalized campaign ID?”
In an early sprint, should a tester validate a data model's fields and allowed values, or focus on business values and end-to-end process behavior?
Asked once · last asked Apr 20 · Scenario · Mid-level
In production, how do you handle schema evolution, such as added or renamed fields and type changes, across ingestion, transformation, and downstream consumers without breaking the pipeline?
Asked once · last asked May 3 · Scenario · Senior
Analytics, Product, Compliance, and Finance define active player, credit earned, and credit redeemed differently, causing inconsistent reporting. How would you define a canonical dataset and a golden path for consumption?
Asked once · last asked May 12 · Scenario · Senior
How would you select and manage security controls for data products and determine the compliance requirements they must meet?
Asked once · last asked Jun 3 · Scenario · Senior
Other ways it was asked
- “What security controls have you managed for data products, and how do you determine the compliance requirements they must meet?”
What does treating data as a product mean in practice, and how would you ensure people follow raw, work, and published zone standards and use a catalog such as Alation?
Asked once · last asked Jun 16 · Scenario · Senior
How would you create and maintain documentation for data pipelines, data models, or data contracts, including runbooks, schema documentation, pipeline diagrams, quality rules, and information needed to operate and maintain the systems?
Asked once · last asked Jun 24 · Scenario · Mid-level
Other ways it was asked
- “How did you create and maintain documentation for your data pipelines, data models, or data contracts? What information did you include, such as runbooks, schema documentation, pipeline diagrams, quality rules, or details to help your team operate and maintain the systems?”
How do you handle schema changes when consuming CDC events into a lake destination?
Asked once · last asked Jul 16 · Scenario · Senior
How would you handle added, removed, or renamed columns when applying CDC events to Delta Lake?
Asked once · last asked Jul 16 · Scenario · Senior
Suppose your final report must include all fields, but some are valid yet disagree. How would you represent those fields to avoid silently inventing a single truth while keeping the report machine-consumable?
Asked once · last asked Jun 30 · Scenario · Senior
How would you handle versioning and schema evolution for a large-scale graph database in Neo4j or AWS Neptune while ensuring data integrity and minimal downtime?
Asked once · last asked Jul 16 · Design · Senior
Related interview questions
- Data Engineer interview questions
- Software Engineer interview questions
- Data Analyst interview questions
- Data Scientist interview questions
- AI Engineer interview questions
- SQL interview questions
- MongoDB and NoSQL interview questions
- ETL and Data Pipelines interview questions
- Spark, PySpark and Databricks interview questions
- Snowflake interview questions
- Azure Data Factory interview questions
- Power BI interview questions
- Excel interview questions
How this list is made
Every question here was asked by a real interviewer in a real job interview that a candidate ran MeetAssist in. After the call, the interviewer's questions are pulled from the transcript. Practice sessions, tests, exams and anything that was not a job interview are dropped.
Questions are rewritten so they make sense on their own: names of people, employers and clients are removed, and speech-to-text mistakes are fixed. Small talk, logistics and questions about one candidate's own résumé are left out. Rephrasings of the same question are merged, and the count is the number of different interviews it came up in.
The interviews come from candidates who use MeetAssist, so they lean towards software, data and AI roles: the shares describe these interviews, not every Data Modeling interview. Answers to the most asked questions are written and checked by an editor; a "common mistake" is shown only when reviews of real answers in these interviews show it, with the count.
274 interviews, 168 candidates, Dec 2025 to Oct 2026. Updated every month with new interviews.
Or add the Chrome extension