Yellow.ai Backend Engineer Interview: Rounds, Questions and Preparation

What the job involves

Backend engineers in Indian SaaS and B2B product companies build and maintain the server-side systems that power multi-tenant applications, APIs, and data pipelines. Day to day, they write code in languages like Java, Python, or Node.js, design database schemas, optimise queries, and ensure high availability for enterprise customers. They also collaborate with product and frontend teams to ship features that handle large volumes of concurrent requests and strict SLAs.

Role and technical knowledge

1. 1

   How would you design a REST API for a multi-tenant SaaS application where each customer's data must be isolated?

   Discuss tenant identification via API keys or JWT, database schema strategies (shared schema with tenant_id vs separate schemas), and access control checks.

2. 2

   Explain the difference between horizontal and vertical scaling. Which one would you prefer for a B2B SaaS backend and why?

   Cover stateless services, load balancers, database read replicas, and the trade-offs of cost and complexity in a growing SaaS product.

3. 3

   What is database indexing? How would you decide which columns to index in a table that stores millions of customer records?

   Explain B-tree indexes, query patterns, cardinality, and the overhead of too many indexes on write performance.

4. 4

   How does a message queue like Kafka help in a SaaS product? Give a use case.

   Mention decoupling services, handling spikes in traffic, asynchronous processing like sending email notifications or generating reports, and ensuring reliability.

Practical scenarios

1. 5

   Your API response time suddenly increases from 200ms to 2 seconds. How would you debug and fix this?

   Talk about checking logs and metrics, profiling slow database queries, identifying N+1 queries, and adding caching or indexes.

2. 6

   A customer reports that their data is missing from a report. How would you investigate and resolve this?

   Outline steps like verifying data ingestion pipelines, checking for failed jobs, reviewing logs, and reproducing the issue with test data.

3. 7

   You need to add a new feature that sends real-time notifications to users. How would you implement it without affecting existing performance?

   Suggest using a message queue with a separate worker service, WebSockets or push notifications, and load testing before rollout.

4. 8

   How would you handle a situation where two services need to update the same database record concurrently?

   Discuss optimistic vs pessimistic locking, database transactions, and using version numbers to avoid lost updates.

Behavioural and HR questions

1. 9

   Tell me about a time you had to learn a new technology quickly for a project. How did you approach it?

   Use the STAR method, highlight self-learning resources, and show how you applied it to deliver results under a deadline.

2. 10

   Describe a situation where you disagreed with a teammate's technical approach. How did you handle it?

   Focus on data-driven discussion, listening to their perspective, and compromising for the project's benefit without ego.

3. 11

   Why do you want to work in a SaaS product company rather than a services company?

   Show interest in building scalable products, working with modern tech stacks, and the opportunity to solve recurring problems for many customers.

4. 12

   How do you prioritise tasks when you have multiple deadlines and limited time?

   Mention using urgency vs importance, communicating with stakeholders, and breaking down tasks to deliver incremental value.

Coding and problem-solving

1. 13

   Write a function to find the first non-repeating character in a string. What is its time and space complexity?

   Use a hash map to count frequencies in one pass, then iterate to find the first with count 1. Aim for O(n) time and O(1) space for fixed character set.

2. 14

   Given a list of API request logs with timestamps, find the busiest 5-minute window. How would you solve this efficiently?

   Sort timestamps, use a sliding window or two-pointer approach, and handle edge cases like ties or empty input.

3. 15

   Implement a simple LRU cache with get and put operations in O(1) time.

   Combine a hash map with a doubly linked list, explain eviction policy, and write clean code with boundary checks.

4. 16

   How would you design a rate limiter for an API endpoint that allows 100 requests per minute per user?

   Discuss token bucket or sliding window algorithms, in-memory vs distributed stores like Redis, and returning 429 status codes.