Data Scientist Interview Questions India 2026: Complete DS Interview Guide

Indian DS interview question types: (1) Statistics and probability ('Explain Type I and Type II errors and their relationship to sample size'; 'What is the p-value and why is a p-value of 0.05 not as meaningful as most people think?'; 'Derive the expected number of coin flips to get heads twice in a row'). (2) Machine learning algorithms ('Compare gradient boosting and random forest: when would you use each and why?'; 'How does XGBoost handle missing values differently from a standard gradient boosting implementation?'; 'What is regularisation and what is the difference between L1 and L2 penalty?'). (3) Python and pandas ('Write a function to compute the rolling 7-day average for a time series with gaps'; 'Find all users who had their first purchase in January but did not purchase in February — using pandas'). (4) SQL ('Using window functions: write a query to find each user's 3rd session'; 'Find the top 10 products by revenue in Q4 that were not in the top 10 in Q3'). (5) A/B testing and experimentation design ('Swiggy wants to test a new checkout flow. Design the experiment: null hypothesis, primary metric, sample size calculation, duration, and what would make you stop early'). (6) ML system design ('Design a fraud detection model for PhonePe UPI transactions').

Statistics topics tested in depth: Probability distributions — know when to apply each: Binomial (binary outcomes: will this user convert? expected failures in 100 trials), Poisson (count events in a fixed interval: orders per hour, support tickets per day), Normal (continuous data: model errors, heights, income — Central Limit Theorem is why it appears so often), Exponential (time between events: user session gaps). Hypothesis testing — understand the mechanics: null and alternative hypothesis, test statistic (t-statistic for means, z-statistic for proportions), p-value (probability of observing your result or more extreme, assuming H0 is true — not the probability that H0 is true), statistical significance (reject H0 when p < alpha). Critically: explain what you lose if you reject H0 with a small sample (underpowered test: high Type II error — you might miss a real effect). A/B testing specifics: power calculation (sample size depends on desired power, alpha, minimum detectable effect, and baseline metric variance), sequential testing (why peeking at results early inflates Type I error), CUPED (Controlled-experiment Using Pre-Experiment Data: variance reduction technique Booking.com open-sourced and widely used at Indian companies).

Senior data scientist interviews at Indian product companies include ML system design: Design a recommendation system (Swiggy or Nykaa product recommendations): cold start problem (new users have no purchase history: use demographic similarity or popular items in their city), collaborative filtering (user-user based on purchase history similarity), matrix factorisation (SVD or ALS for implicit feedback), two-stage approach (candidate generation: fast approximate nearest-neighbour search; ranking: feature-rich gradient boosting model on top candidates), evaluation metrics (offline: NDCG@K, precision@K; online: click-through rate, conversion rate, average order value). Design a fraud detection model (PhonePe UPI): feature engineering (velocity features: transactions per hour for this user, per device, per merchant; deviation features: transaction amount vs user average; graph features: does this merchant have suspicious network connections), model choice (XGBoost for its explainability and handling of imbalanced classes), threshold tuning (precision-recall tradeoff: a false positive blocks a legitimate user, a false negative allows fraud — the business decides the cost ratio), model monitoring (feature distribution drift, score distribution shift, feedback loop when fraud is later confirmed).

Data scientist salary benchmarks: Junior DS (0-2 years): Rs 8-22 LPA at product companies. Mid-level (2-5 years): Rs 22-55 LPA. Senior DS (5-8 years): Rs 50-100 LPA. Principal DS / Head of Data Science: Rs 90-180 LPA. Top DS employers: Flipkart (search ranking, personalisation, fraud: Rs 30-80 LPA mid-senior, one of India's most mature DS environments), Swiggy (demand forecasting, driver supply optimisation, surge: Rs 25-70 LPA), PhonePe (fraud detection, credit scoring: Rs 30-75 LPA), Meesho (social commerce recommendation, seller success ML: Rs 20-55 LPA), Google India (search quality, ads relevance, YouTube recommendations: Rs 60-180 LPA mid-senior, highest-paying DS roles in India), Amazon India (Prime Video recommendations, demand forecasting, Alexa NLP: Rs 45-130 LPA). Analytics consulting firms: Fractal Analytics, Mu Sigma, Tiger Analytics, and ThoughtWorks Data pay Rs 15-40 LPA mid-level — lower base but broader industry exposure across clients.