Expect a mix of coding, design, and applied data scenarios. Use concrete examples from your experience, quantify outcomes, and narrate your trade-offs.

Technical / Domain (Modeling, Warehousing, Governance)

• How would you model a multi-touch marketing attribution dataset, and what are the core dimensions and facts?
• Describe your approach to CDC ingestion and handling late-arriving updates at scale.
• When do you choose star schemas vs. wide tables for BI performance?
• How do you implement data contracts and communicate schema changes across teams?
• Walk through your lineage strategy and how you keep documentation current.

System Design / Architecture

• Design a scalable data platform on Databricks + AWS to serve analytics and ML with shared semantic metrics.
• Architect a CDC→MDW→reverse ETL flow for audience activation with strict SLAs and DQ checks.
• Propose a strategy for cost-optimized, incremental processing of clickstream data.
• How would you introduce streaming for near real-time metrics while keeping batch accuracy?
• Discuss blue/green deploys for data pipelines and how you would validate safety.

Coding / SQL / Python

• Write a SQL query using window functions to compute cohort retention over time.
• Implement a binary search variant and explain time/space complexity and edge cases.
• Given skewed joins in Spark, how do you mitigate skew and optimize performance?
• Write a Python function to mask PII deterministically across datasets.
• Refactor a slow CTE-heavy query for better performance and maintainability.

Problem-Solving / Case Studies

• A key dashboard is stale after a schema change upstream. Walk us through your incident response.
• You have a 1-day SLA to add new attributes to the MDW while maintaining governance. What’s your plan?
• Data volume doubled overnight. How do you triage failures and keep SLAs?
• Your SnapLogic pipeline must enforce transformation guardrails—how do you validate compliance?
• An analyst wants a custom metric that conflicts with the semantic layer. How do you resolve it?

Behavioral / Leadership

• Tell me about a time you led a cross-functional data initiative with ambiguous requirements.
• Describe a time you balanced speed and data quality—what trade-offs did you make?
• How have you mentored others to raise the bar for data engineering practices?
• Share an example where you influenced stakeholders to align on a data roadmap.
• How do you handle on-call data incidents while maintaining project velocity?

Deep Dive into Evaluation Areas

Data Modeling, Warehousing, and Semantic Layers

Intuit relies on well-designed data models to power analytics and AI across product, marketing, and enterprise domains. You’ll be evaluated on dimensional design, semantic layer strategy, schema evolution, and how you enable self-service analytics at scale. Expect to discuss modeling choices in the context of Databricks, AWS data services, dbt, and BI platforms.

Be ready to go over:

• Dimensional vs. data vault vs. wide-table patterns: When to apply each; balancing reusability, performance, and analyst usability.
• Semantic layer design: Translating business logic to governed metrics and entities; lineage and documentation practices.
• Schema evolution and CDC: Handling late-arriving data, SCD types, and upserts at scale.
• Advanced concepts (less common): Knowledge graphs/ontologies, entity resolution, multi-tenant modeling, query acceleration (caching, materializations), cost-aware design on Lakehouse.

Example questions or scenarios:

• “Design a semantic model for customer journeys spanning web, email, and in-product events. How do you define sources of truth and KPIs?”
• “You need to add high-churn attributes to a large fact table. What’s your approach to schema evolution and maintaining SLAs?”
• “Explain when you choose denormalization for BI workloads and how you mitigate data duplication risks.”

Pipelines, Orchestration, and Streaming

Expect scrutiny on ETL/ELT design, orchestration reliability, and observability. You’ll cover dependency management, idempotency, backfills, and cost/latency trade-offs across tools like Airflow, dbt, SnapLogic, Kafka, and reverse ETL patterns.

Be ready to go over:

• Batch vs. streaming: Trigger types, watermarking, late data handling, and exactly-once semantics.
• Orchestration & reliability: DAG design, retries, circuit breakers, SLAs/SLOs, and on-call playbooks.
• Observability & DQ: Parity checks (e.g., CDC→MDW), anomaly detection, data contracts, and incident runbooks.
• Advanced concepts (less common): Change data capture at scale, incremental compaction for Lakehouse, multi-region failover, blue/green data releases.

Example questions or scenarios:

• “Design a CDC-to-warehouse pipeline for marketing activation with strict transformation guardrails. How do you validate parity?”
• “A downstream dashboard is stale. Walk through your triage for lineage, freshness, and schema-change regressions.”
• “Propose a backfill strategy for a 6-month gap without breaching compute budgets.”

SQL, Python, and Applied Coding

You will write and reason about code. While depth varies by team, SQL proficiency and Python for data transformation/automation are standard. Expect to solve problems ranging from analytical SQL to simple algorithms and data manipulation.

Be ready to go over:

• Advanced SQL: Window functions, complex joins, conditional aggregations, performance tuning (partitioning, Z-ordering, statistics).
• Python for data: Pandas vs. Spark, UDF pitfalls, typing and tests, lightweight data quality checks.
• Algorithmic basics: Searching/sorting, complexity, and pragmatic solutions to everyday data tasks.
• Advanced concepts (less common): Vectorized ops vs. UDFs on Spark, Python packaging for ETL repos, CI/CD for data code.

Example questions or scenarios:

• “Given two large tables (events, sessions), compute sessionized KPIs with constraints on runtime and cost.”
• “Implement a binary search variant and discuss edge cases with duplicates.”
• “Write a Python job to apply reversible PII masking before publishing to analytics zones.”