Data Governance, Contracts, and Classification Questions

Governing data at scale: data contracts between producers and consumers, schema evolution/compatibility, data classification and sensitivity tagging, access control, and lineage/cataloging. Covers policy, ownership, and compliance-driven controls over data. The governance layer over the technical stack.

EasyTechnical
49 practiced

Propose a data classification scheme (for example: public, internal, confidential, restricted) for a company's analytics data. For each tier, who should be able to access it and how should it be stored (masking, encryption), and give one example field that would typically land in that tier. Once a dataset is classified, how does that classification concretely drive which technical controls (access, encryption, retention) get applied to it downstream?

HardTechnical
47 practiced

An analyst wants to join a table containing PII (say, user profiles) with an events table but should never see the raw PII columns unless specifically entitled, and in a multi-tenant warehouse a query should never be able to see another tenant's rows even through an intermediate step. Show how you'd structure the SQL (CTEs, views, column masking, row-level security) so that intermediate query steps can't leak PII or cross-tenant data to someone without the right permissions, and describe how you'd test that the protection actually holds.

HardSystem Design
34 practiced

You're responsible for PII controls across a whole analytics toolchain: SQL, a BI tool, object storage, and notebooks. Design the policies and technical controls (masking, tokenization, RBAC, audit logging) plus the automation needed to catch accidental PII leakage, while still letting authorized analysts do their real work in each of those tools.

MediumTechnical
48 practiced

Design a practical, org-wide strategy to detect and mask PII across all your streaming and batch pipelines, not just the ones someone remembered to flag, covering both raw lake data and curated warehouse tables. What detection approaches would you combine (schema tagging, regex pattern matching, ML-based classifiers), what masking or redaction strategy follows once something is found, and how would you handle the inevitable false positives and legitimate exceptions?

MediumTechnical
39 practiced

You're responsible for classifying the sensitivity of columns in a shared CRM dataset (say, contacts and accounts tables) and setting access policy for several internal roles with different needs. Design a classification scheme, a column-level access policy per role, and a masking or tokenization approach for anyone exporting the data. What's the trade-off between tightening this and keeping the sales and analytics teams productive?

Unlock Full Question Bank

Get access to all 6 Data Governance, Contracts, and Classification interview questions and detailed answers.

Sign in to Continue

Join thousands of developers preparing for their dream job.