Tiger Analytics Analyst interview: process, rounds and questions

Written by Mentari as typical of this process — not a record of questions Tiger Analytics has asked. Each comes with the follow-up a good interviewer asks next and what they listen for in the answer. Scenario details are our framing, chosen to fit Tiger Analytics’s sector.

1. 01 Write a function that returns the k most frequent words in a document, and tell me its complexity.

   Follow-up · The document is 20 GB and does not fit in memory. What changes?

   Listens for · A hash map and a heap or sort, stated complexity, and a sensible move to streaming or chunked counting rather than panic.

2. 02 Walk me through the data pipeline of the project you are proudest of — from raw input to the number you reported.

   Follow-up · Which step did you not build yourself, and what would break if it were removed?

   Listens for · Honest ownership and an understanding of the whole system, including the parts a teammate or a library handled.

3. 03 Write a query that gives each store’s month-on-month revenue change.

   Follow-up · A store had no sales in February. What does your query show for March, and is that right?

   Listens for · A window function or self-join chosen deliberately; notices the missing-month gap before being told.

Compiled from 4 candidate reports (2024–2025). Where reports differ, the range is shown rather than one number.

Online assessment — one reported pattern (Data Analyst, December drive) — Another drive reported five sections — aptitude and reasoning, CS basics, then three separate coding problems — without question counts. Reported funnels were steep: roughly 220 took the test and 50 progressed in one drive; 260 sat and 13 progressed in another.^5,6,4

1. Revise probability and statistics to the level of conditional probability, Bayes and expected value, and practise saying them aloud. Every multi-round report names the second technical round as the place difficulty jumped, and it jumped on probability.
2. Do two or three medium coding problems a day with a timer, explaining complexity at the end of each. The online test and the first interview both contain coding; reports say explaining complexity was explicitly rewarded.
3. Write out the data pipeline of your best project and be ready to defend each choice. Projects were the longest part of every interview reported, including what you learned and how you applied it.
4. Prepare one answer each on handling missing values and on approaching a forecasting problem. Both scenarios recur across reports as the analytics-sense questions.
5. Practise working from a hint without restarting. Interviewers are reported to offer hints; what they then watch is whether you can take one.
6. Read what Tiger actually does and decide which kind of team you would want. The HR close reportedly asks exactly that, and a vague answer wastes an easy round.

What Tiger Analytics says it assesses

• "A merit-based recruitment process characterized by rigorous screening"; legitimate openings are posted only on the official careers site and no fee is ever asked¹
• Learning from the first job onward — bootcamps, workshops, hackathons and masterclasses are how the firm describes developing early-career hires²

What candidates report being probed on

• Coding with the interviewer watching, then explaining time complexity^4,5,7
• SQL and DBMS fundamentals, including joins and OOP concepts^5,6
• Conditional probability, Bayes theorem and card or dice questions^5,7
• Projects line by line: what you built, what you learned, how you applied it^4,5,6
• Analytics scenarios — missing values, a sales forecast, a city-traffic problem^4,6

Tiger Analytics interviews like a data-science firm that also expects you to code. Most campus analytics drives test SQL and aptitude; Tiger’s reported rounds add live coding with the interviewer watching, questions about complexity, and then a swing into probability and statistics that several candidates describe as the point where the difficulty jumped. A student who prepared for an "analyst" interview and arrives without having revised Bayes theorem and recursion is the one who reports being caught out.

The second distinctive thing is how much of the process is the online test. Reported funnels are brutal — one drive took 260 students to 13 before anyone was interviewed — and the test is proctored, timed per section and heavy on CS fundamentals, so the interview is reached by a small group who then get an unusually long, exploratory conversation. Interviewers are repeatedly described as patient and willing to give hints; the reports suggest the firm is watching how you use a hint as much as whether you needed one.

Where the sources disagree

The four reports agree on the ingredients and disagree on the count. Two describe four distinct rounds (test, two technical interviews, HR); one describes a test followed by a single combined technical-and-HR conversation; the Trainee Analyst report inserts a résumé-only round between the technical and HR stages. The online assessment is described once in detail (about 32 items in 90 minutes on iMocha) and once as five sections with three coding problems and no counts given, so the 90-minute figure should be treated as one drive’s experience rather than a fixed pattern. No report states an academic cut-off, and Tiger’s own pages publish none; the campus notice is the only authority. Where the official material is silent on process, it is consistent on tone — "rigorous screening" and a heavy emphasis on learning after joining — which matches what candidates report being asked about their interests and the kind of team they want.

4. 04 I toss two fair coins. Given that at least one is heads, what is the probability both are?

   Follow-up · Now I tell you the first coin is heads. Why is the answer different?

   Listens for · Gets one third and one half, and can explain in words why the conditioning differs — the Bayes intuition, not the formula.

5. 05 A column in your dataset has 30% missing values. What do you do?

   Follow-up · The missing values are not random — they are all from one region. Does your answer change?

   Listens for · Asks why the values are missing before choosing a method; recognises that non-random missingness is information.

6. 06 You trained a model with 94% accuracy on a fraud dataset. Is that good?

   Follow-up · Fraud is 2% of transactions. What would you report instead, and to whom?

   Listens for · Class imbalance, precision/recall or cost of errors, and the ability to explain the choice to a non-technical stakeholder.

7. 07 Explain the difference between a process and a thread, using something from your own project.

   Follow-up · Where in that project did the difference actually cost you time?

   Listens for · A definition that is grounded rather than recited; an honest "it did not" is acceptable if it comes with understanding.

8. 08 Here is a puzzle: with 25 horses and a track for 5, find the fastest 3 in the fewest races.

   Follow-up · Now you only need the single fastest. How many races, and why did the answer drop so much?

   Listens for · Reasons aloud, reaches 7 and then 5, and can explain which information the extra races were buying.

9. 09 A city wants to reduce traffic at one junction using data. What would you collect and what would you do with it?

   Follow-up · You can afford one sensor. Where does it go?

   Listens for · Frames the question as a decision, chooses a metric, and makes a constrained choice with a reason.

10. 10 Which kind of team at Tiger would you want to join, and why?

    Follow-up · What have you done in the last six months that shows that preference is real?

    Listens for · Knows the firm does consulting-style analytics work across industries; names a preference backed by something concrete.

You have sales data for 300 stores and need to forecast next month. Where do you start?

Weak · "I would use a machine-learning model like XGBoost on the historical data and tune the hyperparameters to get the best accuracy."

Stronger · "First I would look at the data: how many months, whether some stores are new, where the gaps are and why. Then a simple baseline — last month, or a seasonal average — so I know what a model has to beat. Only then a model, and I would check it per store, because an average error can hide twenty stores forecast badly."

Why the follow-up lands · The interviewer asks "forty stores have three months of history — what do you do with them?" The strong answer already has a place for that question; the weak one has to invent one.

Illustrative — yours must be true.

A Tiger Analytics drive is decided mostly by a proctored test that mixes aptitude, CS fundamentals and coding, with reported shortlists of 5–25% of those who sit it, so the cohort-level lever is coding and fundamentals practice under time pressure well before the drive. Eligibility and invited branches are set per campus and are not published by the company.

For the shortlisted few, Mentari’s data-analyst mock covers the interview that follows — reasoning through SQL and probability aloud, a project deep-dive, and taking a hint without collapsing — and shows the cell who is ready for it. Run Tiger Analytics-pattern mocks for a whole batch.

• Data analyst mock interview
• Software engineer mock interview
• ZS Associates interview
• EXL interview
• Fractal Analytics interview
• Guesstimate Questions: Method + Worked Example
• STAR Method Answers That Survive Follow-ups
• All company guides →

1. Recruitment fraud alert — Tiger Analytics (official) · official · accessed 2026-10-23 · the only official statement about the recruitment process
2. Why join us — Tiger Analytics (official) · official · accessed 2026-10-23
3. Current openings — Tiger Analytics (official) · official · accessed 2026-10-23 · lists roles and locations; no campus process
4. Tiger Analytics interview experience for Data Analyst (on-campus) — GeeksforGeeks · candidate report, 2024-12 · accessed 2026-10-23 · four rounds; 260 sat, 13 advanced
5. Tiger Analytics interview experience (Data Analyst, on-campus) — GeeksforGeeks · candidate report, 2024-12 · accessed 2026-10-23 · 90-minute iMocha test; 220 sat, 50 advanced
6. Tiger Analytics interview experience — Trainee Analyst / Data Science — GeeksforGeeks · candidate report, 2024-12 · accessed 2026-10-23
7. Tiger Analytics interview experience for Data Analyst (on-campus) — 3 — GeeksforGeeks · candidate report, 2025-07 · accessed 2026-10-23 · two rounds; combined technical and HR

Sources checked October 2026 by Sumit Mishra, Mentari founder. Spotted an error, or represent Tiger Analytics? Write to [email protected] — we acknowledge within two working days and correct or remove a disputed statement within 72 hours.

Mentari is not affiliated with Tiger Analytics. This page is pattern-based, compiled from the public sources listed, and is not insider information. Candidate-reported details are reproduced as patterns only, in our own words.