Post
System Design Interview Questions by Company: Google, Meta, Amazon, Uber and More
System design interview questions grouped by company (Google, Meta, Amazon, Microsoft, Apple, Uber, Netflix, Stripe, LinkedIn and more), with what each company's questions tend to test and a link to practise every problem.
By Codemia • November 14, 2026
6 min read
Last updated: November 2026.
Every large tech company asks system design questions, but they do not all ask the same ones. A company's questions tend to look like its own products and its own engineering problems: Google asks about search and storage at enormous scale, Stripe asks about money moving correctly, Uber asks about location and dispatch.
This page groups the system design problems in the Codemia catalogue by the companies they are tagged with, so you can focus your practice on the loop you have coming up. Each problem links to a page where you can design it yourself and get it graded.
Treat it as a practice list, not a list of leaked questions. Interviewers rotate their questions and many problems are asked at several companies. Pick five to eight problems for your target company, do them under a 45 minute timer, and spend the rest of your time on breadth.
Google questions lean towards large scale infrastructure and storage, search, and products used by billions. Expect the interviewer to push on scale estimates and on what happens when a data centre fails.
Good problems to start with:
- URL shortener (easy): ID generation and a read-heavy cache.
- Typeahead suggestion (medium): tries, precomputed top results, and latency budgets.
- YouTube or Netflix (medium): upload pipelines, transcoding and CDNs.
- Google Docs (hard): real-time collaboration and conflict resolution.
- Google Maps (hard): map tiles, routing and live traffic.
- Google Search (hard): crawling, indexing and ranking at web scale.
- Web crawler (hard): politeness, deduplication and a distributed frontier.
- Key-value store (hard): partitioning, replication and consistency choices.
- Distributed file system (hard): chunking, metadata and replica placement.
- Ad click aggregation (hard): stream processing and exactly-once counting.
Meta
Meta questions are about social products: feeds, messaging, comments, likes and media, with very large read volumes and hot keys (a celebrity post, a viral video).
- Twitter (medium): fan-out on write versus fan-out on read.
- Facebook's news feed (hard): feed generation and ranking.
- Facebook Messenger (medium): delivery guarantees, ordering and online presence.
- Instagram (medium): photo storage and the home feed.
- Live comments (medium): pushing updates to millions of viewers.
- Nested comments (easy): tree storage and pagination.
- Video view counter (easy): counting at very high write rates.
- Online presence indicator (easy): heartbeats and fan-out of status.
- Ticketmaster (hard): contention and seat holds. Common at Meta despite not being a social product.
- TikTok (hard): a recommendation-driven video feed.
Amazon
Amazon questions often involve commerce, orders and inventory, and services that must stay up during a traffic spike. Expect follow-ups on availability and on what happens when a dependency is slow.
- E-commerce service (medium): catalogue, cart and checkout.
- Inventory management (medium): stock counts that never oversell.
- Online payment service (medium): idempotency and reconciliation.
- Food delivery service (hard): three-sided marketplace and dispatch.
- Hotel booking service (hard): availability search and double-booking.
- Frequently viewed products (medium): aggregation over a sliding window.
- Top-K request analysis (medium): heavy hitters at scale.
- Task scheduler (medium): durable delayed jobs and retries.
- Uber backend (hard) and Dropbox (hard) are also tagged for Amazon.
Microsoft
Microsoft questions span productivity tools, cloud infrastructure and communication.
- Google Docs (hard): the same collaboration problem as Office online.
- Scalable email service (hard): storage, search and delivery.
- Collaborative meeting scheduler (medium): calendars and free/busy lookups.
- Webhook notification service (medium): retries and delivery guarantees.
- Metrics monitoring and alerting (hard): time series ingestion and alert rules.
- ChatGPT (hard): serving a large language model to many users.
- GPU cluster scheduler (advanced): scheduling ML training jobs.
Apple
Apple questions often involve devices, media and privacy.
- Spotify (medium): music streaming and playlists.
- Podcast hosting platform (medium): media hosting and feeds.
- App distribution platform (medium): an app store.
- Digital wallet (medium): balances that must always be correct.
- Smart home system (hard): device state and commands.
- Multi-device screenshot capture (easy): sync across devices.
- Fitness tracking app (easy): device data ingestion.
Uber
Uber questions centre on location, matching and real-time state.
- Uber backend (hard): driver location updates and rider matching.
- Yelp or nearby friends (hard): geospatial indexes such as geohash and quadtrees.
- Food delivery service (hard): dispatch across three parties.
- Order tracking system (medium): a state machine with live updates.
- Distributed unique ID generator (medium): IDs without coordination.
- Distributed locking system (medium): leases and fencing tokens.
- Task scheduler (medium).
Netflix
- YouTube or Netflix (medium): encoding and adaptive streaming.
- Global content distribution network (hard): edge caching worldwide.
- Live video streaming platform (medium).
- E-commerce recommendation system (medium): the same candidate-and-rank pattern as content recommendations.
Stripe, PayPal and Shopify
Payments and commerce companies care most about correctness: no double charges, no lost money, a clear audit trail.
- Online payment service (medium).
- Digital wallet (medium).
- API rate limiter (medium).
- Webhook notification service (medium).
- Scheduled digital transactions (hard).
- Flash sale system (medium).
- Shopify (medium) and online coupon service (medium).
LinkedIn and Twitter (X)
- Twitter search (medium): inverted indexes over a stream of posts.
- Graph search for a social network (hard): queries over connections.
- Distributed counter (medium): likes and views.
- Platform like Reddit (medium): voting and ranking.
- Push notification service (hard).
DoorDash, Airbnb and Dropbox
- Airbnb: hotel booking, event lifecycle management.
The problems that come up everywhere
Some problems are tagged for almost every large company, because they test core ideas any interviewer can probe:
If you only have a week, do these first. How to answer 10 classic system design questions outlines the key decision in each.
How to prepare for a specific company
- Pick five to eight problems from the company's list above, mixing one easy, several medium and a couple of hard.
- Do each under a 45 minute timer before reading any solution, then compare against the editorial and note what you missed.
- Think about the company's own product. If you are interviewing at Uber, be ready to talk about geospatial indexing whatever the question.
- Rehearse out loud. The AI interviewer runs any problem in the catalogue as a spoken mock and grades it against that problem's key points.
- Do not skip breadth. You may well get a question that is not on any list. System design practice problems, from easy to hard covers the patterns.
For a full schedule, see how to practise system design: a 4-week plan.