Most-asked design questions

This is a list of the high-level design questions that come up most often in interviews, grouped by the kind of system they describe. For each one you get the two or three ideas the question is really testing, a rough sense of how often it is asked, and links to the lessons and labs in this course that teach those ideas.

How to use this list

Don't try to memorise 50 designs. There aren't 50 different answers here. There are perhaps a dozen core ideas (fan-out, caching, partitioning, idempotency, queues, geospatial indexes, consensus, stream processing...), and every question is a different mix of them. The "What it really tests" column is the point of this page. When you can look at a new question and name its two or three hard parts within a minute, you are ready.

New to system design? Finish Part 1 of the path first, then do these five, in this order: a URL shortener (lab), a rate limiter (lab), a news feed (lab), a chat app (lab) and a flash sale (lab). Between them they cover most of the core ideas below, from caching and IDs to fan-out, real-time delivery and contention.

Then a good way to work through the rest:

1. Start with the High rows. They are the ones most likely to come up and they cover most of the core ideas.
2. Pick one question per category and design it end to end in 45 minutes, out loud, using the interview framework. Then open the linked lab and compare.
3. For the rest, just do the first 10 minutes: requirements, rough numbers, and naming the hard parts. Check that your hard parts match the column.
4. Notice the repeats. "Ticket booking", "flash sale" and "hotel booking" are the same problem in different clothes: many people want one scarce thing at the same instant.

Frequency is a rough judgement, not a statistic. High means it shows up at most companies that run design rounds. Med means common but not universal. Low means it shows up mainly at companies whose product is close to the question.

Social and feeds

Feed and social questions are favourites at consumer-internet companies, FAANG included. They all revolve around one tension: a post is written once and read by thousands, so do the work when it is written or when it is read?

Messaging and real-time

Anything with "live", "chat" or "presence" is about long-lived connections and getting a message to the one server that holds the right connection. Messaging-heavy products (WhatsApp, Slack, Discord) and Indian apps with live order tracking or live cricket ask these a lot.

Media, storage and files

These are about big blobs: where they live, how they reach users quickly, and how you avoid sending the same bytes twice.

Commerce and payments

Money and inventory questions are where correctness beats speed. They are a staple at e-commerce and fintech companies; in India expect them at Flipkart, Razorpay, PhonePe, CRED and friends. Listen for "never twice": never charge twice, never sell the same seat twice.

Location and delivery

Anything with a map is a geospatial indexing question plus a stream of location updates. Food delivery and quick commerce questions are especially common at Swiggy, Zomato, Zepto and Uber.

Infrastructure building blocks

These questions ask you to build one of the boxes you normally just draw. They test whether you know what is inside. Infrastructure-heavy companies and senior rounds lean on them.

Search and discovery

Data and analytics

Counting things at scale: clicks, views, metrics. The hard parts are almost always duplicates, late data and approximate answers that are good enough.