Data Engineering System Design: Interview Roadmap

System design rounds test whether you can turn vague requirements into a data platform that works, survives failure and fits a budget. Preparation is mostly practice with a repeatable structure.

How to practise each case study

1. Read only the problem statement and requirements.
2. Set a 40-minute timer and design out loud, drawing the flow.
3. Compare with the case study: what did you miss? Which trade-offs did you not name?
4. Answer the follow-up questions at the end of the case study.
5. Repeat the same case a week later.

What interviewers look for

• Clarifying questions before designing.
• Rough numbers that drive choices.
• Correct handling of reruns, late data and duplicates.
• Explicit trade-offs tied to requirements.
• Awareness of operations, security and cost.

The plan

1. 01

   Stage 1: Learn the framework

   Clarify requirements, estimate scale, sketch end to end, go deep on risks, cover reliability, security and cost, then trade-offs.

   Typical effort

   1 week

   Outcome

   You can run the framework from memory on any prompt.

   • Data Engineering System Design Cheat Sheet
2. 02

   Stage 2: Know the building blocks

   Ingestion, storage layers, table formats, batch and stream processing, orchestration, serving.

   Typical effort

   2 weeks

   Outcome

   You can explain when to use each block and its failure modes.

   • ETL vs ELT and Modern Data Pipelines
   • Data Lakes, Lakehouse Architecture and Delta Lake
   • Kafka and Real-Time Data Engineering
3. 03

   Stage 3: Master reliability patterns

   Idempotency, late data, duplicates, schema evolution, quality gates, observability and backfills.

   Typical effort

   1–2 weeks

   Outcome

   You can explain how your design survives reruns, late data and bad data.

   • Idempotency in Data Pipelines
   • Data Quality: Checks, Contracts and Failure Handling
   • Data Pipeline Observability Fundamentals
4. 04

   Stage 4: Practise batch designs

   Batch pipeline, cloud warehouse and reporting platform case studies.

   Typical effort

   1 week

   Outcome

   You can design and defend each in 40 minutes.

   • Design a Batch Ingestion Framework
   • Design a Cloud Data Warehouse for a SaaS Product
   • Design an Analytics and BI Platform
5. 05

   Stage 5: Practise streaming and CDC designs

   Real-time analytics, Kafka ingestion, clickstream and CDC case studies.

   Typical effort

   1–2 weeks

   Outcome

   You can handle event time, ordering and delivery-semantics follow-ups.

   • Design a Real-Time Analytics Pipeline
   • Design a Real-Time Streaming Platform
   • Design a Clickstream Analytics Pipeline
   • Design a CDC Pipeline from an OLTP Database to the Warehouse
6. 06

   Stage 6: Practise a platform design

   Company-wide lakehouse with governance and many teams.

   Typical effort

   1 week

   Outcome

   You can discuss organisation, governance and cost, not just pipelines.

   • Design a Lakehouse with Bronze, Silver and Gold Layers
7. 07

   Stage 7: Rehearse trade-offs out loud

   Practise naming options, choosing for the requirement and stating the cost.

   Typical effort

   Ongoing

   Outcome

   Your answers sound like decisions, not lists.

   • Data Engineering Interview Trade-offs: A Practical Guide