🎯 Interview & Career · Chapter 1

How data engineering interviews are structured

A data engineering interview isn't one exam. It's five or six different exams graded by different people on different things — and most rejections come from preparing for the wrong one.

Study this lesson free in the app →

The basics

A data engineering interview loop is the sequence of rounds a company uses to decide whether to hire you: usually a screen, an online test, two or three technical rounds, and a manager/HR conversation.

Why it matters: each round measures something different. The SQL round measures whether you can turn a business question into a correct query. The design round measures whether you can make and defend trade-offs. The behavioural round measures whether people want to work with you. Prepare only for SQL and you'll pass two rounds and fail the loop.

The mental model — a typical loop (it varies by company):

Recruiter / HR screen — 15–30 min: experience, notice period, expected CTC
Online assessment — 60–90 min: SQL, Python, sometimes aptitude MCQs
Technical round(s) — SQL live, Python/data manipulation, Spark or warehouse concepts
Design round — pipeline or data-model design (more weight for 3+ years' experience)
Project deep-dive / hiring manager — your past work, decisions, ownership
HR / offer — compensation, joining date

Words to know: loop, screen, OA (online assessment), bar, debrief, take-home.

Explained

Priya had solved 300 SQL problems. She cleared the online test at Meghalaya Cabs with 100%, sailed through the SQL round, and was rejected.

The feedback: “Design round: jumped to tools, no numbers. Project round: couldn't explain why the pipeline was built that way.”

She'd prepared brilliantly for two rounds out of five.

The driving test

A driving licence test in India isn't one test. There's the written test on signs and rules, the ‘8’ track for control, the road test with traffic, and the examiner watching whether you check mirrors without being told.

Someone who memorised every road sign can still fail on the ‘8’. Someone who drives perfectly can fail the written.

Each part is scored separately and you need all of them. An interview loop works exactly the same way — and like the examiner, interviewers notice habits (narrating, checking edge cases) as much as results.

A typical loop, and what each round scores

Smaller companies merge rounds (SQL + Python in one hour). Large ones may add a bar-raiser or a second design round. Always ask the recruiter for the exact structure — it's a normal question and they usually answer it.

Three kinds of company, three kinds of loop

IT services (large consultancies) — high-volume hiring, especially for freshers: an online test with aptitude, verbal and coding sections, then one technical and one HR round. Breadth of fundamentals matters more than depth.

Product companies and startups — fewer candidates, deeper rounds: live SQL with follow-ups, Python data manipulation, a design round, sometimes a take-home assignment. Depth and reasoning matter most.

Global capability centres (GCCs) — in-house tech centres of multinational companies (banks, retailers, pharma) in Bengaluru, Hyderabad, Pune, Chennai. Often structured like product companies, with more emphasis on cloud platforms they use (Snowflake, Databricks, AWS/Azure) and on stakeholder communication.

The job description is your best clue: tools listed under ‘must have’ usually become a round.

What changes with experience

                 fresher / 0–2 yrs     3–5 yrs              6+ yrs
────────────────────────────────────────────────────────────────────────
OA / aptitude    often, filters        sometimes            rare
SQL              core: joins, groups   windows, perf,       assumed; quick
                                       tricky edge cases
Python           basics, lists/dicts   pandas/PySpark,      code quality,
                                       testing              design of code
Design           rare or light         one pipeline design  decides the level
Project round    college/intern work   ownership, numbers   influence, trade-offs,
                                                            mentoring
Behavioural      light                 STAR stories         conflict, leadership

The senior loop isn't harder SQL. It's more weight on judgement: design, trade-offs, and how you made other people effective.

What happens after you leave

Every interviewer writes feedback within a day: a rating (often strong no / no / yes / strong yes) plus evidence. “Wrote a correct window query for top-3 drivers per city; handled ties after a hint; didn't consider NULL fares.”

In the debrief, the panel and hiring manager read all of it together. Patterns they look for:

Consistent signals across rounds (‘clear communicator’ in three rounds)
Any strong no with concrete evidence — often decisive
Level fit: is this a strong mid-level or a weak senior?

That's why a brilliant SQL round doesn't rescue a design round where you couldn't defend a choice.

Gotchas

Preparing only for SQL. It's the most visible round, not the only one.
Not asking about the structure. “How many rounds, and what does each cover?” is a normal question for a recruiter.
Treating the screen as a formality. Vague answers about notice period or expectations can end it there.
Silence in live rounds. A correct answer in silence scores lower than a slightly slower one explained aloud.
Underestimating the project round. It's where they test whether your CV is true.
Taking a ‘no’ personally as a verdict on your ability. It's one panel's read on one day; get feedback when offered and adjust.

Five exams, five graders, one decision.

Screen, test, SQL, Python, design, project, HR — ask which ones you'll face, and prepare for each on its own terms.

Words to know

Interview loop	The full sequence of rounds a company uses for one hiring decision.
Online assessment (OA)	A timed, auto-graded test — SQL, Python, sometimes aptitude — used to filter at volume.
Debrief	The meeting where interviewers combine per-round feedback into a hire/no-hire and level decision.
Take-home	An assignment done in your own time, usually followed by a discussion round.
Bar raiser	An interviewer from outside the team whose role is to keep the hiring standard consistent.
GCC	Global capability centre — a multinational's in-house technology centre in India.

In two minutes

A loop is typically 4–6 rounds over 2–6 weeks; each round scores a different skill.
Screen checks fit and logistics; the online assessment filters on SQL and Python basics at volume.
Live technical rounds check whether you can reason out loud, not just whether you reach the answer.
Design and data-modelling rounds grow in weight with seniority; for 4+ years they often decide the level.
The project deep-dive tests whether you really owned your work: decisions, numbers, failures.
Service companies, product companies and GCCs run different loops — read the JD and ask the recruiter.
Interviewers write feedback per round and a debrief combines them; one strong ‘no’ can outweigh several weak ‘yes’es.

Test yourself

1. Which round is most likely to decide the level for a candidate with 6+ years' experience?

The design / data-modelling round
The online aptitude test
The recruiter screen
Basic SQL syntax

Show the answer

The design / data-modelling round — Senior candidates are expected to know SQL and Python. Judgement — design trade-offs, scale, failure handling — is what separates mid-level from senior.

2. What does the recruiter screen usually cover?

Live window-function coding
System design
Experience summary, notice period, current and expected CTC
Spark internals

Show the answer

Experience summary, notice period, current and expected CTC — It's a logistics and fit filter. Being clear and consistent here avoids being dropped before any technical round.

3. In a live SQL round, a candidate gets the right answer in silence. Another reaches it slightly slower, narrating assumptions and edge cases. Typically who scores higher?

The silent one — correctness is all that matters
Neither — speed is the only criterion
Both equal
The narrating one — interviewers grade reasoning they can hear

Show the answer

The narrating one — interviewers grade reasoning they can hear — Interviewers must write evidence. Narration gives them evidence of reasoning, edge-case thinking and communication.

4. A JD lists ‘Must have: Snowflake, dbt, Airflow’. What should you infer?

Nothing — JDs are generic
Expect questions on warehouse modelling, dbt tests/models and DAG design
Only SQL will be asked
Python won't be asked

Show the answer

Expect questions on warehouse modelling, dbt tests/models and DAG design — Must-have tools often become a round or a big part of one. Prepare concrete examples using them.

5. You cleared SQL and Python rounds strongly but got a ‘strong no’ in design. Most likely outcome?

Often rejected or down-levelled — a strong no with evidence carries heavy weight
Hired — two out of three is enough
Automatically hired at a lower salary
Another SQL round

Show the answer

Often rejected or down-levelled — a strong no with evidence carries heavy weight — Debriefs look for consistent signal. A well-evidenced strong no is hard to outvote, though some companies may offer a lower level.

6. Trap: is it rude to ask the recruiter how many rounds there are and what each covers?

Yes — it signals you're gaming the process
Only after an offer
No — it's a normal, expected question
Only for senior roles

Show the answer

No — it's a normal, expected question — Recruiters are measured on successful hires and routinely share the structure. Not asking just means preparing blind.

There are 8 questions, 3 puzzles and 5 recall cards for this lesson in the app — the recall cards come back on later days, just before you would forget.

Learn it so it stays. Read it, test yourself, then recall it on two different days. The app schedules that for you, free.

Start this lesson →