Skip to main content
🗄️

Data Engineer Interview Prep — pipelines, warehouses, SQL — for engineers in Tashkent and Central Asia

Data engineering interviews are deeper on SQL than backend interviews and lighter on coding than ML interviews. Practice the questions Tashkent fintechs, telcos, and EU remote employers actually ask, with real-time AI scoring on communication, technical depth, problem-solving, and cultural fit.

Data engineering in Tashkent is hired most by fintechs (Click, Payme, Uzum), telcos (Beeline, Ucell), and a growing number of EU remote employers building data platforms in Snowflake or BigQuery. The interview format is consistent: deep SQL, one pipeline design question, one debugging-a-broken-job question, and the same behavioral set as any backend role.

Below are the questions that come up most. Practice them out loud — pipelines and schemas are easier to draw than to talk about cleanly under pressure.

Core skills tested

  • Advanced SQL: window functions, CTEs, query optimization
  • One pipeline framework: Airflow, Dagster, or dbt
  • One warehouse: BigQuery, Snowflake, or Redshift
  • Python for data work (pandas, pyarrow, basic OOP)
  • Data modeling: star vs snowflake schema, slowly changing dimensions
  • Streaming basics: Kafka or Pub/Sub fundamentals
  • Data quality: tests, monitoring, lineage
  • Cost awareness: knowing why a query is expensive

Salary ranges in Tashkent (2026)

Approximate. Remote-first European roles typically pay 30–50% above local rates.

Junior

9–14M UZS / month

Mid-level

20–32M UZS / month

Senior

38M+ UZS / month (or EUR remote)

What you will actually be asked

Pulled from real interviews recorded on NextSuhbat. Each item is a question you should expect, plus what the interviewer is really testing.

  1. 1

    Recruiter screen

    Walk me through the most complex pipeline you have built.

    Why it is asked: Two minutes. Source, transforms, destination, schedule, and the metric you owned. Be specific about volumes.

  2. 2

    Technical

    Write a SQL query to find the second-highest salary in a table without using LIMIT.

    Why it is asked: Window-function classic. Tests if you reach for DENSE_RANK or a self-join.

  3. 3

    Technical

    Explain the difference between a star schema and a snowflake schema, and when you would pick each.

    Why it is asked: Tests dimensional modeling fundamentals. Tie each choice to query patterns and ETL cost.

  4. 4

    Technical

    What is a slowly changing dimension and how do you handle Type 2?

    Why it is asked: Bar for any warehouse role. Cover effective_from / effective_to and the join pattern.

  5. 5

    Technical

    A nightly job is now running 4 hours instead of 30 minutes. How do you investigate?

    Why it is asked: Practical debugging. Cover query plan changes, data volume growth, partitioning, skew, and an obvious recent deploy.

  6. 6

    Coding

    Given two tables — orders(user_id, ts, amount) and users(id, country) — write SQL for monthly revenue per country, with month-over-month change.

    Why it is asked: Window functions, joins, date truncation in one query. Real production task.

  7. 7

    Coding

    Deduplicate a stream of events that may arrive out of order, with a 24-hour late tolerance.

    Why it is asked: Tests if you understand watermarks and idempotent writes.

  8. 8

    System design

    Design a pipeline that ingests Click-like payment events, deduplicates, joins reference data, and serves a daily dashboard.

    Why it is asked: Cover ingestion (Kafka or batch), staging vs marts, late data, and dashboard freshness SLA.

  9. 9

    Behavioral

    Tell me about a time data quality cost the business something visible.

    Why it is asked: Specific number, specific incident, specific change you made. Vague answers signal you have not been close to prod.

  10. 10

    Behavioral

    How do you decide between writing a one-off SQL query and building a proper pipeline?

    Why it is asked: Tests judgment. Frequency of use, fan-out of consumers, and reproducibility are the right axes.

Practice these questions out loud — for free

Reading is not practice. Run a 20-minute AI mock interview in English, Russian, or Uzbek and get a scorecard against communication, technical depth, problem-solving, and cultural fit.

Start free mock interview

Built in Tashkent for Central Asia. All practice sessions support English, Russian, and Uzbek voice.

Related reading