Question 1

When would you choose a Snowflake schema over a Star schema?

Accepted Answer

Star: One fact, denormalized dimensions—simple, fewer joins, fast. Snowflake: Normalized dimensions (e.g., dim_product → dim_category → dim_category_group)—more joins, less redundancy. Why Snowflake: When dimension tables are large and shared—a single dim_category serves many products; denormalizing would duplicate millions of rows. Avoids inconsistent attributes across copies (e.g., category name updated in one place). Storage: Snowflake saves space when hierarchy is deep and wide....

Question 2

Deadlock: Definition and necessary conditions

Accepted Answer

**Conditions**: Mutual exclusion, hold-and-wait, no preemption, circular wait. **Prevention**: Break circular wait by ordering resources. **Detection**: DBs detect and resolve. **Best practice**: Design to avoid; use timeouts.

Question 3

Describe your approach to the case study.

Accepted Answer

**Situation**: Case studies often present ambiguous business problems with undefined constraints and multiple valid solutions.

**Task**: Deliver a structured, defensible solution aligned with business goals while demonstrating technical depth and communication.

**Action**: (1) Understand—clarify success metrics, constraints, data availability, and stakeholders; (2) Design—propose architecture (batch vs....

Question 4

HTTP vs HTTPS Protocol

Accepted Answer

**HTTP**: Unencrypted; port 80. **HTTPS**: TLS/SSL; port 443; confidentiality, integrity, server auth. Prevents eavesdropping, MITM. Search engines favor HTTPS; browsers mark HTTP insecure. Use HTTPS everywhere for production APIs. Certificates: Let's Encrypt, cloud provider....

Question 5

How did you structure your solution?

Accepted Answer

**Framework**: (1) Problem definition—goals, constraints, users; (2) Architecture—components, flow, tech with rationale; (3) Design—models, APIs, failure modes; (4) Implementation—phased, risks; (5) Validation—criteria, testing. Use diagrams (data flow, ER); ADRs for decisions....

Question 6

IPv4 vs IPv6

Accepted Answer

IPv4: 32-bit (~4.3B addresses); exhausted. IPv6: 128-bit (3.4×10^38); no NAT, built-in IPsec, stateless auto-config. WHY IT MATTERS: IoT and global scale need IPv6; legacy systems and internal pipelines often IPv4-only. SCALABILITY: Dual-stack (AWS, GCP) allows gradual migration; IPv6 simplifies addressing for massive device counts. COST: No direct cost difference; IPv6 reduces NAT complexity and can lower operational overhead....

Question 7

Light Bulb Problem: Three switches outside a room control three bulbs inside. Determine which switch controls which bulb with minimal trials.

Accepted Answer

Use heat as a second observable. Turn switch 1 ON for 10 min, OFF. Turn switch 2 ON. Enter room. Bulb OFF and warm = switch 1; ON = switch 2; OFF and cold = switch 3. Minimal trials: one entry. WHY: Systematic thinking—leverage multiple dimensions (light, heat) to disambiguate with minimal actions....

Question 8

Present your findings and recommendations.

Accepted Answer

Structure: (1) Executive summary—key insights, top 3 recommendations. (2) Context—problem, sources, methodology. (3) Analysis—visualizations, stats; avoid raw tables. (4) Findings—segmented by dimension. (5) Recommendations—actionable, impact/effort prioritized. (6) Next steps—timeline, owners. WHY: Audience-appropriate; technical detail in appendix. Use clear visuals, annotate anomalies, link recommendations to findings....

Question 9

RAM Working Principles

Accepted Answer

RAM: Volatile, byte-addressable storage; faster than SSD; CPU accesses via memory controller; data in cache lines. DRAM (capacitors) vs. SRAM (flip-flops). WHY FOR DE: Spark, Pandas rely on memory; exceeding it causes spill and GC pressure. SCALABILITY: Cluster memory limits parallelism; monitor OOM and GC....

Question 10

Semaphore in OS and its Use Case

Accepted Answer

Semaphore: Count-based sync; P (wait) decrements, blocks if 0; V (signal) increments. Binary (0/1) = mutex. USE: Limit concurrent DB connections, producer-consumer buffers, rate limiting. WHY: Mutex is one-at-a-time; semaphore allows N concurrent....

ZS Associates Data Engineer Interview Questions

Companies asking these questions

All 23 Questions

More Interview Prep Guides

Get All Answers in PDF Format

ZS Associates Data Engineer Interview Questions

Companies asking these questions

All 23 Questions

More Interview Prep Guides

Get All Answers in PDF Format