Question 1

Conflict Resolution - how you handled project conflicts with teammates

Accepted Answer

Situation: A teammate insisted on full refresh for a large fact table; I advocated incremental loads for cost and speed. The disagreement escalated and risked project timelines. Task: Resolve the technical conflict without damaging the relationship. Action: I scheduled a 1:1, shared benchmarks (incremental 5x faster, ~80% cost savings, same correctness under our SCD2 contract). I proposed a pilot on one table and offered to own implementation and validation....

Question 2

Missed Timelines - instance of missing delivery timelines and how you communicated and delivered

Accepted Answer

STAR: SITUATION—Regulatory deadline for data pipeline; mid-sprint schema mismatch would delay delivery. TASK—Meet deadline while minimizing risk. ACTION—Escalated immediately to PM and compliance; proposed phased delivery—MVP (core fields) by deadline, full scope two weeks later; documented risks and got sign-off; delivered MVP on time; completed full scope in follow-up sprint. RESULT—Zero compliance breach; stakeholder trust maintained. COMMUNICATION: Transparent, early escalation....

Question 3

What would you do if a critical data pipeline failed during a holiday?

Accepted Answer

STAR: SITUATION—P1 on holiday. TASK—Restore. ACTION—Acknowledge; triage impact/root cause; mitigate (rollback/hotfix); communicate ETA; fix; retro after. Example: On-call; source API change; temp filter to unblock; full fix next day; added validation....

Question 4

Anagram Detection - find all anagrams from a given list of strings

Accepted Answer

**Logic:** Anagrams = same char counts. **Approach:** Group by sorted string or char-frequency tuple. **Python:** `groups = defaultdict(list); [groups[tuple(sorted(Counter(s).items()))].append(s) for s in words]`. O(n·k log k) for n words, k max length. **Alternative:** Sort each as key. **Why:** Deduplication, fuzzy matching....

Question 5

Find pairs with sum X from a list of numbers

Accepted Answer

**Hash set:** For each num, if (X-num) in seen, count pair. Add num to seen. O(n). **Duplicates:** Counter. **Why:** One pass. **Production:** Handle overflow; duplicates.

Question 6

Reverse a Linked List - implement solution for singly linked list

Accepted Answer

**Why In-Place Reversal:** O(n) time, O(1) space. Recursive version uses O(n) stack—fails for long lists (stack overflow).

**Algorithm:** Iterate, reverse next pointers: save next, point current to prev, advance prev and curr. Handles empty and single-node.

**Analogy in Data Eng:** Reverse order of events (LIFO processing), undo stacks....

Question 7

SQL Question - Cricket Scoreboard Table - aggregations, joins, grouping

Accepted Answer

Cricket scoreboard: tables for players, matches, innings, balls. Aggregations: runs per player (SUM), wickets (COUNT), strike rate (runs/balls*100). Joins: players to innings to balls. Group by match, player. Example: SELECT player_id, SUM(runs), COUNT(CASE WHEN wicket THEN 1 END) FROM balls GROUP BY player_id, match_id. **Why it matters**: Design choices compound at scale—wrong approach can cause 100× overhead....

Question 8

SQL Question - Inner Join with Null Handling, compute total row count

Accepted Answer

INNER JOIN with NULL: NULL keys don't match (NULL = NULL is false), so they're excluded. Count: SELECT COUNT(*) FROM a INNER JOIN b ON a.key = b.key. If key has nulls, inner join count is less than min(|a|,|b|). To include null matches: ON (a.key = b.key OR (a.key IS NULL AND b.key IS NULL)). Count depends on null handling. **Why it matters**: Design choices compound at scale—wrong approach can cause 100× overhead....

Question 9

Apache Spark Fundamentals - failures, job optimization, resource utilization

Accepted Answer

**Why this matters at scale**: Failures and misconfig cost time and money; optimization compounds across thousands of jobs. **Failures & resolutions**: Executor OOM—increase memory, reduce partition size, or optimize (avoid broad joins). Driver OOM—avoid collect(); use checkpointing for long lineage. Stragglers—speculative execution, salting for skew. **Optimization**: Partition pruning, predicate pushdown, broadcast joins, AQE....

Question 10

PySpark Coding Challenge - dataset with 4-5 columns, solve data processing problem on paper

Accepted Answer

**Why it matters**: At scale, design choices directly impact reliability, latency, and cost. Wrong decisions compound across jobs and teams.

PySpark on-paper challenge: Read data, identify columns, state transformations (filter, join, agg), write pseudocode or actual PySpark. Example: Given id, amount, date—compute daily sum per id. `df.groupBy('id','date').sum('amount')`....

Paytm Data Engineer Interview Questions

Companies asking these questions

All 11 Questions

More Interview Prep Guides

Get All Answers in PDF Format

Paytm Data Engineer Interview Questions

Companies asking these questions

All 11 Questions

More Interview Prep Guides

Get All Answers in PDF Format