Interview Prep · ML Systems Lab

Razorpay Senior Data Scientist Interview Guide

Razorpay's Senior Data Scientist loop is distinct from the Applied Scientist track. DS roles emphasise A/B testing, analytics, marketplace experimentation, and business judgment. The loop is shorter than the MLE loop but probes experimentation more deeply. This guide walks through both.

Razorpay runs separate tracks for Applied Scientist (ML-heavy, production engineering) and Senior Data Scientist (analytics-heavy, experimentation, business intelligence). This guide covers the DS track, which differs from the MLE track in important ways.

The loop structure (Senior Data Scientist, L5)

Standard loop: recruiter screen → SQL + Python coding → product analytics case → A/B testing + experimentation deep dive → hiring manager.

The experimentation deep dive is the highest-weight round.

Round-by-round breakdown

Round 1 — Recruiter screen. Standard.

Round 2 — SQL + Python coding. 60 minutes. Heavy SQL emphasis (more than the MLE track), data manipulation in pandas, business logic implementation. Razorpay tests SQL deeper than most Indian unicorns for DS roles.

Round 3 — Product analytics case. 75 minutes. Real scenario: "merchant churn dropped 8% last month, then bounced back 3%. Investigate and report what's likely driving it." Tests the discipline of metric decomposition, cohort analysis, and hypothesis-driven investigation.

Round 4 — A/B testing + experimentation. 60 minutes. Deep probing on: power analysis, sample size, multiple testing corrections, SRM, peeking, novelty effects, marketplace SUTVA violations, sequential testing. The round explicitly tests sophistication on experimentation, not just basics.

Round 5 — Hiring manager. 45 minutes.

What Razorpay DS weights distinctively (vs MLE track)

1. SQL depth. Razorpay DS interviews probe SQL much more than the MLE track. 2. Business judgment. DS candidates are expected to translate analytical findings into business recommendations with clear trade-offs. 3. Marketplace experimentation. Razorpay's two-sided market (merchants + payment users) creates SUTVA-violation concerns. DS candidates are expected to articulate these.

Top 10 questions Razorpay Senior DS candidates face

1. "Merchant churn dropped 8% last month. Investigate." 2. "Design an A/B test for a new merchant onboarding flow. What's your sample size?" 3. "Your experiment shows +5% conversion but +20% support tickets. Do you ship?" 4. "Walk through power analysis for a 1% expected lift on a 10% conversion metric." 5. "Marketplace SUTVA violation: your treatment merchants got more transactions, but those came from control merchants. How do you measure the real treatment effect?" 6. "SRM check fires on an A/B test. What do you do?" 7. "Sequential testing vs fixed sample size. When each?" 8. "How would you measure the LTV of a new merchant cohort?" 9. "Walk through the most impactful analysis you've delivered." 10. "How do you handle multiple comparisons when running 20 metrics on one experiment?"

The prep path through MSL

For a Razorpay Senior DS loop:

  • Tier 1 (Statistics & Estimation) — all 4 posts, particularly Post 113 (Hypothesis Testing).
  • Tier 5 (Evaluation & Diagnostics) — Post 131 (Error Analysis) and Post 42 (Offline ≠ Online).
  • Cross-link to PAL — Razorpay DS roles are partly PAL territory; PAL's experimentation depth covers what MSL doesn't.
  • For practice: CausalInferenceTab (experiment design, SUTVA), MockInterview with a Razorpay DS JD pasted.

    Common failure modes

  • Treating it like the MLE track (under-investing in SQL and experimentation).
  • Treating A/B testing as basic (Razorpay probes deep).
  • Inability to discuss marketplace dynamics for experimentation.
  • Compensation

    Razorpay Senior Data Scientist in 2026 ranges roughly ₹35 lakh – ₹65 lakh for 5-7 YOE. Staff DS reaches ₹85 lakh+. The DS track typically pays slightly less than the MLE track but has different career trajectories.

    Continue interactively
    Read this post inside ML Systems Lab — with Simplify toggle, interview Q&As, inline glossary, and the MLE Path forward pointer.
    Open in MSL →