Product Company · Gurugram
Zomato interviews test algorithmic thinking, backend development skills, and cultural alignment for engineering and analytics roles.
Process: Online Coding → Technical × 2 → Cultural Fit
What does Zomato's data platform look like and what kind of data roles do they hire for?
Tip: Zomato hires Data Analysts, Data Engineers, and ML Engineers. Their data platform processes order data, restaurant ratings, search clicks, and delivery telemetry. Roles require SQL proficiency, Python, and familiarity with Spark or Kafka. Analytics interviews include case studies on restaurant churn or user cohort analysis.
Write a SQL query to find restaurants with a rating drop of more than 0.5 compared to the previous month.
Tip: Use LAG() window function: LAG(avg_rating, 1) OVER (PARTITION BY restaurant_id ORDER BY month). Filter where current_month_rating - lag_rating < -0.5. Zomato analytics rounds are SQL-heavy — window functions appear in almost every interview.
How would you design Zomato's restaurant discovery and ranking algorithm?
Tip: Ranking signals: distance, rating (Bayesian average, not raw), delivery time, cuisine preference (personalised), restaurant freshness (new listings), sponsored listings. Serving: pre-compute scores per geo-cell per user-segment, refresh every 30 min. Real-time: adjust for current wait time and weather.
Explain the difference between OLTP and OLAP databases. Which does Zomato need for real-time orders vs analytics?
Tip: OLTP (Online Transaction Processing): many small reads/writes, row-oriented, MySQL/Postgres — for live order state. OLAP (Online Analytical Processing): large aggregations, column-oriented, Redshift/BigQuery/ClickHouse — for analytics. Zomato runs both: OLTP for operations, OLAP warehouse for business intelligence.
Write a Python function to calculate the rolling 7-day average for a time series of daily orders.
Tip: Use pandas: df["orders"].rolling(7).mean(). From scratch: sliding window with a deque, sum the last 7 values, divide by min(window_size, i+1). Zomato DA interviews test pandas fluency — also know resample() for time-based aggregation.
How would you identify and handle a sudden spike in restaurant cancellations on Zomato's platform?
Tip: Detection: real-time dashboard with cancellation rate per restaurant/zone alerts when >3σ above baseline. Investigation: join with order events (accepted but late? restaurant marked busy?). Action: auto-pause high-cancellation restaurants, alert account manager, surface to user as "temporarily unavailable."
What is A/B testing and how would you design an experiment to test a new restaurant ranking model?
Tip: A/B test: random split users into control (current ranking) and treatment (new model). Metrics: primary (order conversion rate), guardrail (cancellation rate, delivery time). Duration: run until statistical significance (p < 0.05). Pitfalls: novelty effect, SUTVA violations (users share restaurants).
Describe a time you used data to change the direction of a project or team decision.
Tip: Use STAR. Zomato values data-driven decisions. Show: you identified a metric that contradicted intuition, surfaced it clearly to stakeholders, proposed an alternative direction, and measured the outcome. Quantify the impact — 'increased CTR by 12%' is stronger than 'improved performance.'
What is the difference between precision and recall? In which Zomato use case does each matter more?
Tip: Precision: of items flagged positive, what fraction are actually positive. Recall: of all actual positives, what fraction did you catch. Fraud detection: high recall (catch all fraud, tolerate false positives). Restaurant recommendation: high precision (show only great matches, missing some is OK).
How does Zomato use real-time data to optimise delivery partner allocation?
Tip: This is an assignment problem: match delivery partners to orders to minimise total expected delivery time. Real-time constraints: partner is moving, order might not be ready, multiple orders can be batched. Zomato's approach: combinatorial optimisation with a 30-second re-solve window using auction-based algorithms.
Zomato Limited interviews follow a 3-round process. Here is what to expect and how to prepare for each stage.
Upload your resume and get questions scored across technical depth, communication, structure, confidence, and relevance — the same criteria Zomato panels use.