Fractal Analytics Imagineer interview: process, rounds and questions

Written by Mentari as typical of this process — not a record of questions Fractal 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 Fractal Analytics’s sector.

1. 01 Introduce yourself technically — the stack you have actually shipped with, not everything you have touched.

   Follow-up · Pick the API or library on your résumé you understand least. What does it do under the hood?

   Listens for · Honest scoping; some understanding of the tools they claim.

2. 02 Write a query that returns each customer’s most recent order and the days since their previous one.

   Follow-up · What does it return for a customer with only one order?

   Listens for · Window functions used deliberately; edge case noticed before being told.

3. 03 When would you choose a non-relational database over a relational one — in your own project?

   Follow-up · Your project used MongoDB. What query became harder because of that choice?

   Listens for · A trade-off grounded in their own system, not a textbook list.

Compiled from 7 candidate reports (2024–2026) and the official hiring page. Where reports differ, the range is shown rather than one number.

Online assessment — three reported configurations — The 2026 VIT drive reports a roughly 90-minute multi-section test with aptitude, Python, Java, SQL and programming questions; a 2025 data-science drive reports 13 questions on DoSelect (11 MCQs and two coding-based data-science tasks); an older Imagineer account reports 70 questions in 75 minutes across data analysis, reasoning, quantitative and verbal sections. Fractal describes its platform as AI-proctored with domain-specific question banks.^5,6,8,9

1. Write SQL daily until joins, grouping, subqueries and window functions come without thinking — then practise writing them while explaining aloud. SQL is tested three times: as full queries in the assessment, live in the first interview, and conceptually.
2. Prepare every résumé project to implementation depth: which API, which library, what failed. Both 2026 accounts describe the interview going into APIs and technologies named on the résumé.
3. Do timed aptitude and data-interpretation sets, and be ready for an adaptive test. One drive’s assessment got harder after correct answers; sectional pacing matters.
4. Practise Python output-prediction questions and two or three small programs. Python basics appear in the assessment and the technical interview across drives.
5. For a data-science drive, rehearse one end-to-end prediction task with evaluation metrics explained. The 2025 data-science assessment used a wine-quality prediction and model evaluation question.
6. Decide your answers on relocation, higher studies and the five-year plan. All three are reported in the HR round; hesitation is what interviewers notice.

What Fractal Analytics says it assesses

• Imagineer is "the Imagineer Program for campus hiring", recognised by the Training APEX Awards for "transforming graduate hires"^1,2
• Fractal’s Iqigai recruiting platform runs data-science and technology assessments from an "unlimited question bank" with domain-specific customisation and AI proctoring³

What candidates report being probed on

• Writing SQL queries live, and explaining relational versus non-relational databases^5,6,7
• Every technology and API named on the résumé, with implementation detail^5,6
• Basic Python programs and output prediction^7,8
• Logic puzzles such as three switches and three bulbs⁷
• Model evaluation on an end-to-end prediction task (data-science drives)⁸
• Honesty about the résumé, relocation, higher studies, five-year plans^7,5

Fractal’s process is the most SQL-centred of the analytics drives in these guides. SQL appears in the online assessment as full queries to write, again in the first technical interview as a live exercise on a shared screen, and again conceptually as relational-versus-NoSQL questions. The 2026 candidates’ advice is blunt: master SQL and know your projects. A student who can only describe joins, rather than write one while someone watches, is reported to stall in the first interview.

The second distinctive feature is the assessment itself. Fractal builds and sells an AI-proctored assessment platform, and its campus test shows it: candidates report an adaptive version where questions get harder after correct answers, a version that required recording a spoken explanation of an aptitude answer, and a data-science version with end-to-end prediction tasks. Expect the test to be unlike the generic aptitude papers used by IT-services recruiters, and expect proctoring.

Where the sources disagree

The 2026 VIT accounts and the 2024–25 accounts agree on the sequence — assessment, technical interviews, HR — and disagree on the test. One drive reports roughly 90 minutes of mixed aptitude and technical questions including Java; another reports 13 questions with two coding tasks on DoSelect; an older Imagineer report describes 70 questions in 75 minutes with a verbal section. The count of technical interviews is one in the 2024 intern drive and two in the 2026 Imagineer drive. The official sources confirm the programme and the platform but not the pattern, which is consistent with a company that tunes its own assessment per drive. Eligibility is reported as 7.0 CGPA in most 2026 reposts and 8.0 in one campus, so it is set per campus; the Workday requisition for Campus Imagineer Hiring 2026 is the closest thing to an official notice and should be read for the current drive.

4. 04 Predict the output of this Python snippet with a mutable default argument.

   Follow-up · Why does Python behave that way, and how would you write it safely?

   Listens for · Understands evaluation-time defaults; proposes None-and-assign.

5. 05 Three switches outside a room control three bulbs inside; you may enter once. Which switch controls which bulb?

   Follow-up · Now there are four bulbs. Does your method still work?

   Listens for · Reasons aloud, uses a non-visual property, and tests the method against the new constraint.

6. 06 Your model predicted wine quality with 0.62 R². Is that good?

   Follow-up · Which single feature would you drop first, and how would you decide?

   Listens for · Context for the metric; a principled feature-selection answer rather than "try everything".

7. 07 Normalise this order table to third normal form, then tell me when you would deliberately denormalise.

   Follow-up · Your analytics dashboard is slow. Which join do you remove, and what do you risk?

   Listens for · Knows the forms; knows denormalisation is a performance trade with an integrity cost.

8. 08 What do you know about what Fractal does, and which client problem would you want to work on?

   Follow-up · Why that one over the others?

   Listens for · Knows Fractal does AI and analytics for consumer, healthcare and financial clients; a specific preference.

9. 09 Smart work or hard work?

   Follow-up · Give me an example from your degree where one of them failed you.

   Listens for · A real example rather than the platitude the question invites.

10. 10 Are you planning higher studies in the next two years?

    Follow-up · If a GATE result came through in March, what would you do?

    Listens for · A consistent, decided answer.

Write a query that returns each customer’s most recent order and the gap in days since their previous one.

Weak · Writes a GROUP BY with MAX(order_date), then stalls on the "previous order" part and says it would need a second query.

Stronger · Uses a window: LAG(order_date) OVER (PARTITION BY customer_id ORDER BY order_date), computes the difference, then filters to the latest row per customer with ROW_NUMBER — and says aloud what happens to customers with a single order (NULL gap).

Why the follow-up lands · The interviewer asks "what does your query show for a customer with one order?" The strong answer already handled it; the weak one has to rebuild.

Illustrative — yours must be true.

A Fractal Imagineer drive filters on a proctored, sometimes adaptive, 90-minute test heavy on SQL and Python, then interviews shortlisted students with a live SQL exercise and a résumé deep-dive. Cohort preparation is SQL query writing and timed data-interpretation practice; the CGPA cut-off (7.0 in most reports, 8.0 at one campus) is set per drive.

Mentari’s data-analyst mock rehearses the interview half — SQL reasoning aloud, project depth, a puzzle — for shortlisted students before the drive. Run Fractal Analytics-pattern mocks for a whole batch.

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

1. Annual report — strategy (campus hiring through the Imagineer Program) — Fractal (official) · official · accessed 2026-10-23
2. Training APEX Awards recognise Fractal’s Imagineer Program — Fractal Analytics Academy (official) · official · accessed 2026-10-23
3. Iqigai — recruiting and assessment platform — Fractal (official) · official · accessed 2026-10-23 · describes the assessment platform, not the Imagineer test
4. Campus Imagineer Hiring 2026 — requisition — Fractal careers, Workday (official) · official · accessed 2026-10-23
5. Fractal Analytics Imagineer interview experience — on-campus, VIT Chennai 2026 — GeeksforGeeks · candidate report, 2026-08 · accessed 2026-10-23
6. Fractal Imagineer interview experience (VIT, 2026) — OAHelper · candidate report, 2026-08 · accessed 2026-10-23 · one Python problem and three SQL questions in the assessment
7. Fractal interview experience — on-campus, two-month project intern — GeeksforGeeks · candidate report, 2024-09 · accessed 2026-10-23 · adaptive assessment; 20-minute technical and 15-minute HR interviews
8. Fractal Analytics interview experience for Data Science — GeeksforGeeks · candidate report, 2025-06 · accessed 2026-10-23 · DoSelect test: 11 MCQs and two data-science coding tasks
9. Fractal Analytics interview experience for Imagineer — GeeksforGeeks · candidate report, 2024-10 · accessed 2026-10-23 · 70 questions in 75 minutes; two technical interviews
10. Fractal Analytics interview experience — 2 — GeeksforGeeks · candidate report, 2024-11 · accessed 2026-10-23 · older Imagineer drive republished 2024; cited for the two-interview structure
11. Fractal Imagineer hiring 2026 — eligibility repost — freshercareers.in · candidate report · accessed 2026-10-23 · job-board repost of the 2026 drive criteria

Verified October 2026 by Sumit Mishra, Mentari founder. Spotted an error, or represent Fractal 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 Fractal 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.