Swiggy
Leading Indian food delivery aggregator with complex real-time logistics.
4 Rounds
~15 Days
Hard
The Interview Loop
Recruiter Screen (30 min)
Standard fit check, behavioral questions, and resume overview.
Technical Loop (3-4 Rounds)
Deep dive into domain knowledge, coding, and system design.
Interview Question Bank
Machine Learning Engineer
•
Behavioral
•
medium
Tell me about a time a machine learning model you deployed failed or degraded in production. How did you identify the issue and resolve it?
#Ownership
#Post-mortems
#Problem Solving
#Incident Management
Machine Learning Engineer
•
Behavioral
•
medium
At Swiggy, Product Managers often request multiple ML features simultaneously. How do you prioritize which model to build first?
#Stakeholder Management
#Agile
#Product Sense
#Prioritization
Machine Learning Engineer
•
Behavioral
•
medium
Describe a situation where you had to explain a complex ML model's decisions to non-technical stakeholders, like City Operations Managers.
#Communication
#Empathy
#Cross-functional
#Explainable AI
Machine Learning Engineer
•
Behavioral
•
hard
Tell me about a time you optimized an existing ML pipeline to save cloud costs or significantly reduce inference latency.
#Engineering Excellence
#Cost Optimization
#Latency
#System Architecture
Machine Learning Engineer
•
Behavioral
•
easy
Why do you want to join Swiggy? What specific challenges in the food delivery or quick commerce space excite you from an ML perspective?
#Domain Interest
#Motivation
#Company Knowledge
Machine Learning Engineer
•
Coding
•
medium
Write a SQL query to calculate the 7-day rolling average of delivery times for each delivery zone.
#Window Functions
#Date Functions
#Time Series
Machine Learning Engineer
•
Coding
•
medium
Write a Python function to compute the NDCG@K given a list of predicted scores and a list of true relevance labels.
#Python
#Array Manipulation
#Math
Machine Learning Engineer
•
Coding
•
hard
A delivery partner needs to pick up orders from 2 restaurants and deliver them to 2 customers. Write an algorithm to find the shortest valid route (pickups must precede respective drop-offs).
#Graphs
#Shortest Path
#Backtracking
#TSP
Machine Learning Engineer
•
Coding
•
medium
Implement a Trie data structure to support search autocomplete for Swiggy Instamart items. It should support insert, search, and return top 3 matching prefixes.
#Trees
#String Manipulation
#Tries
#Design
Machine Learning Engineer
•
Coding
•
medium
Given an 'orders' table, write a SQL query to find the top 3 most ordered dishes per city in the last 30 days.
#Window Functions
#Aggregation
#Joins
Machine Learning Engineer
•
Coding
•
medium
Given a delivery bag with a maximum weight capacity W, and a list of orders with specific weights and delivery payouts, write a function to maximize the payout.
#Dynamic Programming
#Optimization
#Knapsack
Machine Learning Engineer
•
Coding
•
medium
Implement the K-Means clustering algorithm from scratch in Python using NumPy. Do not use scikit-learn.
#Machine Learning Algorithms
#Python
#Math
#NumPy
Machine Learning Engineer
•
Coding
•
easy
Write a Python script using Pandas to perform stratified sampling on a dataset of 10 million Swiggy orders, stratified by 'city_id'.
#Python
#Pandas
#Data Processing
Machine Learning Engineer
•
Coding
•
medium
Given a list of active time intervals for a delivery executive (e.g., [[10, 12], [11, 15], [16, 18]]), write a function to merge overlapping intervals to calculate total active hours.
#Arrays
#Sorting
#Intervals
Machine Learning Engineer
•
System Design
•
hard
Design Swiggy's ETA (Estimated Time of Arrival) prediction system from scratch. How would you account for food preparation time, delivery partner assignment, and real-time traffic?
#ML System Design
#Regression
#Real-time Systems
#Geospatial Data
Machine Learning Engineer
•
System Design
•
hard
Design a personalized restaurant recommendation system for the Swiggy homepage. How do you balance relevance, novelty, and business metrics like delivery cost?
#Recommender Systems
#Collaborative Filtering
#Two-Tower Models
#Ranking
Machine Learning Engineer
•
System Design
•
hard
How would you design a dynamic 'surge pricing' model for Swiggy delivery during peak hours or bad weather conditions?
#Dynamic Pricing
#Elasticity
#Regression
#Marketplace Matching
Machine Learning Engineer
•
System Design
•
medium
Design a machine learning system to detect promo-code abuse and fraudulent orders in real-time.
#Anomaly Detection
#Classification
#Graph ML
#Real-time Inference
Machine Learning Engineer
•
System Design
•
medium
How would you design a model to predict the exact food preparation time for a specific dish at a specific restaurant?
#Time Series
#Regression
#Feature Engineering
Machine Learning Engineer
•
System Design
•
medium
Design a system to predict Delivery Partner churn. How would you define 'churn' for gig workers who don't have formal contracts?
#Survival Analysis
#Classification
#Churn Prediction
#Feature Engineering
Machine Learning Engineer
•
System Design
•
medium
Design an intent classification model for Swiggy's customer support chatbot. How do you handle out-of-domain queries and multi-lingual text?
#NLP
#Classification
#Real-time Inference
#Chatbots
Machine Learning Engineer
•
System Design
•
hard
Design a demand forecasting system for Swiggy Instamart dark stores to predict inventory requirements for the next 7 days.
#Time Series Forecasting
#Supply Chain
#Demand Prediction
#Deep Learning
Machine Learning Engineer
•
Technical
•
medium
Explain the difference between pointwise, pairwise, and listwise learning to rank. Which approach would you use for Swiggy's dish search, and why?
#Learning to Rank
#Information Retrieval
#NLP
Machine Learning Engineer
•
Technical
•
medium
How do you handle the cold start problem for a newly onboarded restaurant on Swiggy that has no historical order data?
#Recommender Systems
#Cold Start
#Content-based Filtering
#Heuristics
Machine Learning Engineer
•
Technical
•
easy
What is the difference between NDCG and MAP? In what scenario would you prefer one over the other for evaluating search results?
#Evaluation Metrics
#Recommender Systems
#Search
Machine Learning Engineer
•
Technical
•
medium
Why is XGBoost often preferred over Random Forest for tabular data tasks like ETA prediction? Explain the mathematical difference in how they build trees.
#Tree Models
#Ensemble Methods
#Gradient Boosting
Machine Learning Engineer
•
Technical
•
medium
How do you detect and handle data drift in a production ML model, such as our ETA prediction model during a sudden lockdown?
#Model Monitoring
#Concept Drift
#Statistical Tests
Machine Learning Engineer
•
Technical
•
medium
If an offline recommendation model shows a 10% improvement in NDCG, but the online A/B test shows a drop in conversion rate, what could be the reasons?
#A/B Testing
#Evaluation Metrics
#Offline-Online Discrepancy
Machine Learning Engineer
•
Technical
•
easy
In our fraud detection dataset, only 0.1% of transactions are fraudulent. How do you train a robust model on this highly imbalanced data?
#Classification
#Imbalanced Data
#Loss Functions
Machine Learning Engineer
•
Technical
•
hard
Explain the architecture of a Two-Tower neural network. How would you use it to match users to restaurants, and how do you serve it efficiently at scale?
#Neural Networks
#Embeddings
#Retrieval
#Vector Databases
Machine Learning Engineer
•
Technical
•
hard
How would you build an ML pipeline to map millions of unstructured menu items from different restaurants into a standardized Swiggy dish taxonomy?
#Clustering
#Embeddings
#LLMs
#Text Classification
Machine Learning Engineer
•
Technical
•
hard
You have trained a heavy PyTorch model for image quality assessment of restaurant food photos. How do you optimize it for low-latency inference on AWS?
#Model Quantization
#ONNX
#TensorRT
#Caching
#Model Serving
Machine Learning Engineer
•
Technical
•
medium
What is the difference between AUC-ROC and PR-AUC? Why is PR-AUC generally preferred for evaluating models on highly skewed datasets?
#Evaluation Metrics
#Probability
#Imbalanced Data
Machine Learning Engineer
•
Technical
•
medium
Explain Matrix Factorization in the context of collaborative filtering. What are its main limitations when applied to a fast-changing catalog like Swiggy Instamart?
#Recommender Systems
#Linear Algebra
#ALS
#Collaborative Filtering
Machine Learning Engineer
•
Technical
•
hard
How would you use Graph Neural Networks (GNNs) to improve the routing efficiency of delivery partners in a complex road network?
#GNNs
#Spatial Data
#Routing
#Advanced ML
Difficulty Radar
Based on recent AI-sourced data.
Meet Your Interviewers
The "Standard" Interviewer
Senior EngineerFocuses on core competencies, system constraints, and clear communication.
SimulateUnwritten Rules
Think Out Loud
Always explain your thought process before writing code or drawing architecture.