Design a URL Shortener
bit.ly at scale: turn long URLs into short codes, redirect in milliseconds from anywhere, count the clicks.
The Cursor for system design interviews
Stop studying system design. Start practicing it.
Practicing system design interviews with ChatGPT is like coding by copy-paste: every question means explaining your design again.
Here the interviewer, your whiteboard and the AI Mentor share one room, so you can go straight to the deep dives, where interviews are actually decided.
No sign-up. No credit card. Start in one click.
Chat 1: the interviewer
Interviewer
Traffic spikes 10x and your click queue starts backing up. What happens?
Chat 2: my questions
Chat 2 can’t see the interview. You re-explain your design every time.
Reference designs35
Deep dives187
Questions from 52 companies
Why this exists
The diagram was never the problem. The deep dives were.
- 01
Onsite after onsite, the same round
I started interviewing again and kept making it to onsites. Every time, system design is where it fell apart. Rejection after rejection.
- 02
Prep stopped at the diagram
Most material ends at load balancer, cache, database, done. My high-level designs were fine. I got stuck when the interviewer asked how exactly I would handle a hot partition, or what happens when this queue backs up.
- 03
Two AI windows
So I practiced with AI: one chat interviewed me, and I asked my follow-up questions in a second chat so the interview stayed clean. But the second chat had no idea what we had discussed. I was re-explaining my design every time.
- 04
So I put it all in one room
It felt like pasting code into ChatGPT before the model moved inside the editor, where it could see everything. So I built myself a tool that keeps the interview, the whiteboard and the AI Mentor in one place. It turned into this site, and I still use it.
ChatGPT in a tabSysDesignPrep.com
Your whiteboardCan’t see it. You describe it in text.The AI Mentor reads the board as you draw
The interview so farOnly what you paste in, againEvery turn, without re-explaining
Follow-up questionsDerail the interview, or go to a chat that knows nothingAsked beside the interview, which stays clean
Past the diagramOnly as deep as you know to ask187 written deep dives: hot keys, backpressure, failure modes
What strong looks likeYou would have to know alreadyA strong candidate’s answer, turn by turn, then a scored debrief against it
How it works
Two ways to prepare, in one room
A whiteboard, the conversation between interviewer and candidate, and an AI Mentor beside it. First you walk through it, then you sit it.
Walkthrough
Walk through a strong candidate's answer
- A recorded interview, played a turn at a time
- The whiteboard fills in as they talk
- You answer first at the key decisions
- Ask the AI Mentor anything along the way
Mock interview
Then sit the same question yourself
- An AI interviewer that probes what you say
- It sees what you draw on the board
- Timed, or untimed with the AI Mentor beside you
- A scored debrief, stage by stage
- 01
The interviewer asks
A real system design question, opened the way a real interviewer opens it: the problem in two sentences, then "where would you like to start?"
- 02
A strong candidate answers
Clarifying questions, requirements with numbers, estimates, the API, then the design, drawn on the whiteboard as they talk, one bottleneck at a time. You press Next for each turn.
- 03
You answer first at the key decisions
Which database? How do you generate the codes? 301 or 302? The walkthrough pauses and asks what you would say, then shows what the candidate said.
- 04
Ask the AI Mentor anything
Why Postgres and not Cassandra? What is request coalescing? Would my idea work? The AI Mentor sees the interview and the whiteboard where you are, and answers there and then.
- 05
Then sit it yourself
The same question as a mock interview: an AI interviewer asks, probes what you say and what you draw, and debriefs you, stage by stage, against the walkthrough you went through.
Questions
Each one teaches patterns the others do not
Every question comes three ways: a walkthrough to go through, a mock interview to sit, and a full written reference design to read.
Design Slack
Real-time team messaging: channels, DMs, presence, history and search for millions of concurrent users.
Design a Payment System
Accept card payments for an e-commerce platform, move money correctly exactly once, and be able to prove it.
Design a News Feed
The home timeline of Twitter, Instagram or Facebook: posts from people you follow, ranked, fresh, and fast for hundreds of millions of users.
Design YouTube
Upload, transcode, and stream video to a billion viewers: a processing pipeline on one side, a CDN problem on the other.
Design Google Maps
Map tiles, place search, turn-by-turn routing with live traffic, and ETAs for a billion users.
Design a Matchmaking System
Group online game players into fair, low-latency matches within seconds, at millions of concurrent players.
Design Uber
Match riders with nearby drivers in seconds, track trips live, and price them, for millions of concurrent users worldwide.
Design a Notification System
Deliver push, email, SMS and in-app notifications to hundreds of millions of users: the right channel, once, on time, and never at 3 a.m.
Design Google Docs
Real-time collaborative editing: many cursors in one document, every keystroke converging to the same text, offline edits merged on reconnect.
Design an LLM Inference Service
Serve a large language model behind an API: queue the requests, batch them onto scarce GPUs, stream tokens back, and keep every accelerator busy without blowing the latency budget.
Design a Distributed Cache
A Redis-like cache as a service: hundreds of nodes, terabytes of memory, sub-millisecond gets, and a cluster that survives losing a machine mid-request.
Design a Web Crawler
Fetch billions of pages a month politely, discover new links, skip duplicates, and keep the copy fresh without getting the crawler banned.
Design a Rate Limiter
A distributed service that decides, in under a millisecond, whether a request from a client is allowed under its quota.
Design Dropbox
Keep a folder identical on every device: upload huge files in resumable chunks, push only what changed, and reconcile two machines that both edited the same file offline.
Design Ticketmaster
Sell 60 000 seats in ten minutes to a million people at once: hold a seat while someone pays, never sell it twice, and keep the site up when demand is a hundred times supply.
Design Search Autocomplete
Suggest the top ten completions after every keystroke, in under fifty milliseconds, from what the world searched for in the last hour, without suggesting anything that gets you in the newspaper.
Design an Ad Click Aggregator
Count a million ad clicks a second, show advertisers a dashboard that is seconds behind, stop a campaign the moment its budget runs out, and be exactly right when the invoice is cut.
Design a Distributed Job Scheduler
Run millions of cron jobs and one-off tasks on time across a fleet, exactly once each, and keep going when a worker dies halfway through one.
Design a Stock Exchange
Match buy and sell orders in microseconds, in a strict and provable order, never lose a fill, and stay correct when the matching engine dies mid-trade.
Design a Project-to-Contractor Matching System
Match posted projects with qualified, available contractors across a 500 k-person talent pool, and learn from who actually gets hired.
Design Instagram
Upload photos, publish posts and stories, and load a feed of them for 500 M daily users, with every image served from a CDN in milliseconds.
Design WhatsApp
Deliver 50 billion end-to-end encrypted messages a day to phones that are often offline, with sent, delivered and read ticks, and a server that cannot read any of it.
Design Netflix
Stream a fixed catalog to 40 M people at once, 200 Tbps at peak, starting in under two seconds, from cache servers placed inside the viewers’ own internet providers.
Design Yelp
Find the best places near you, filtered by what is open now, from 200 M businesses, in under 200 ms, and keep ratings honest as reviews arrive.
Design a Real-Time Leaderboard
Rank 300 M players of a mobile game in real time: the global top 100, your exact rank among millions, your friends, and daily, weekly and season boards that reset on time.
Design a Distributed Key-Value Store
A Dynamo-style store that keeps 10 billion keys on dozens of machines, survives node and rack failures, and lets each caller trade consistency for latency with N, R and W.
Design a Unique ID Generator
Hand out a million 64-bit ids a second across many hosts and regions, never twice, roughly in time order, with no central bottleneck and no trust in the wall clock.
Design LeetCode
Run untrusted code from strangers against hidden tests in a dozen languages, return a fair verdict in seconds, and survive 50,000 people submitting in the last minutes of a contest.
Design Airbnb
Search 7 M listings by place, dates and price, book a stay without ever selling the same night twice, and keep calendars in sync with hosts’ other platforms.
Design a Distributed Message Queue
A Kafka-style log that ingests a million messages a second, keeps them in order per key, replicates them so no acknowledged message is lost, and lets many consumer groups read at their own pace.
Design a Metrics and Monitoring System
Ingest 10 million samples a second from 100,000 hosts, keep them queryable for a year at shrinking resolution, draw dashboards in under a second, and page the right person within a minute.
Design Tinder
Show each person a deck of nearby profiles that fit both people’s preferences, never repeat one, take 2 billion swipes a day, and announce a match the instant two people like each other.
Design Reddit
Serve communities of posts with deeply nested comment threads, take tens of thousands of votes a second, rank by hot and best, and keep a 50,000-comment thread fast while it is still growing.
Design DoorDash
Take a food order, get a restaurant to cook it and a courier to arrive just as it is ready, and keep three parties in sync across millions of orders a day.
Study guides
All 183 guides The theory behind the questions
One page per building block, written to be said out loud in an interview. Free to read.
01How to run a system design interview02Common system design interview mistakes03"What interviewers expect at mid, senior and staff levels"04System design patterns cheat sheet05"Conway's law and team boundaries"06Back-of-envelope estimation07Back-of-envelope estimation, worked examples08Latency numbers every engineer should know09Queueing theory and capacity planning10Scalability fundamentals11"Stateful vs stateless services"12What happens when you type a URL13Networking for system design14Global traffic management, DNS and anycast15Tail latency, percentiles and queueing16"Designing read-heavy vs write-heavy systems"17Compression in system design18Load balancing19Autoscaling and capacity planning20Kubernetes for system design interviews21Serverless architecture22Caching23Cache invalidation strategies24Redis data structures and patterns25Skip lists and sorted sets26"Redis vs Memcached"27CDNs and edge computing28HTTP caching and cache headers29Edge computing30Object storage and large files31Distributed file systems and erasure coding32Storing billions of small files (Haystack)33Delta sync and content-defined chunking34Media uploads and processing pipelines35Video streaming: transcoding, HLS, DASH and adaptive bitrate36Live video streaming37Video calls and WebRTC38Distributing software updates at scale39Choosing a database40Document databases and MongoDB-style modelling41Database indexing: B-trees, LSM trees and secondary indexes42Storage engines, write-ahead logs and durability43Transactions, isolation levels and locking44"ACID vs BASE"45"Optimistic vs pessimistic locking"46Scaling a relational database47Online schema migrations and data backfills48Database normalization for system design49Data modelling for reads50Data modelling for Cassandra and DynamoDB51Storing and querying graphs52Large-scale graph processing53Modelling comments and nested threads54Sharding and partitioning55Consistent hashing56Hot keys, hot partitions and data skew57Geospatial indexing and proximity search58"Geohash vs quadtree vs S2 vs H3"59Routing and shortest paths at scale60Geocoding and place search61Location tracking at scale62Dispatch and marketplace matching63Surge and dynamic pricing systems64Replication and consistency65"Consistency models: strong vs eventual and everything between"66Quorums and leaderless replication67Merkle trees68The CAP theorem and PACELC69Search, indexing and autocomplete70How Elasticsearch works71Search ranking and relevance72Web search engine architecture73Tries and autocomplete74Bloom filters, HyperLogLog and count-min sketch75Bloom filters explained76Vector search, embeddings and RAG77Approximate nearest neighbour indexes (HNSW, IVF, PQ)78Consensus, leases and coordination79How Raft consensus works80Gossip protocols and failure detection81Spanner, TrueTime and distributed SQL82Distributed locks, leases and fencing tokens83Multi-region architecture84"Active-active vs active-passive"85Cell-based architecture86Message queues and event streams87How Kafka works88"Kafka vs RabbitMQ vs SQS"89"Synchronous vs asynchronous communication"90Event-driven architecture91"At-most-once, at-least-once and exactly-once delivery"92Event sourcing, CQRS and change data capture93Change data capture and the outbox pattern94Background jobs and task queues95Delayed jobs, timers and distributed cron96Workflow orchestration and durable execution97Batch and stream processing98"Lambda vs Kappa architecture"99MapReduce and Spark100Windowing, event time and watermarks101Web crawling at scale102OLTP versus OLAP: analytics, columnar storage and data warehouses103Dimensional modelling and star schemas104Data lakes, warehouses and lakehouses105Data quality and data contracts106Recommendation systems107Feature stores and serving machine learning models108Multi-armed bandits and exploration109Voting and ranking algorithms110Ad serving and real-time auctions111Time-series data112Counting at scale113API design114"REST vs gRPC vs GraphQL"115API gateways and backends for frontends116Serialization formats and schema evolution117Pagination: offset, cursor and keyset118Bulk imports and exports119Designing webhooks120Microservices, service discovery and API gateways121Migrating legacy systems with the strangler fig pattern122Service discovery and service mesh123Multi-tenancy124Authentication, authorisation and data protection125"JWT vs session cookies"126OAuth 2.0 and OpenID Connect explained127Authorization and permission systems128Encryption and key management129End-to-end encryption in messaging130"Privacy in system design: deletion, retention and residency"131Audit logs and tamper-evident records132Trust and safety, moderation and abuse prevention133Running untrusted code safely134Real-time systems: WebSockets, SSE and push135"WebSockets vs Server-Sent Events vs long polling"136Presence and managing millions of connections137Group chat and channel fan-out138"Fan-out on write vs fan-out on read"139Activity feeds and notification inboxes140Push notifications141Sending email at scale142Designing an email inbox (Gmail-style)143IoT device fleets and telemetry144Offline-first apps and sync145"CRDTs vs operational transformation"146Mobile system design147Matchmaking and skill rating148Multiplayer game servers149Rate limiting and resilience150The thundering herd problem151Rate limiting algorithms152DDoS protection and abusive traffic153Load shedding and backpressure154Distributed transactions and idempotency155Idempotency keys in practice156Inventory, reservations and flash sales157Bookings, reservations and availability158Calendar and appointment scheduling systems159Shopping cart and checkout design160Money in system design: ledgers, idempotency and reconciliation161Digital wallets and stored balances162Real-time fraud detection163Low-latency systems and matching engines164Subscriptions and recurring billing165Real-time market data distribution166Unique ids, ordering and time167"Lamport clocks, vector clocks and hybrid logical clocks"168Message ordering and sequence numbers169Hashing and encoding for system design170Time zones, dates and calendars in system design171Internationalization and localization at scale172Observability, operations and rollouts173Distributed tracing and structured logging174SLIs, SLOs and error budgets175Deployment strategies and safe releases176Feature flags and A/B testing177Cost-aware system design178Testing distributed systems and chaos engineering179Incident response and postmortems180Backups and disaster recovery181Building LLM features: serving, latency and cost182Serving LLMs on GPUs183Designing AI agent systems
Pricing
Start free. Go Pro when it helps.
The URL Shortener walkthrough is free with no account, and signing in opens every walkthrough. Every plan adds the AI mock interviews and the AI Mentor, including what is added after you buy.
Monthly
$49 / month
billed monthly, cancel any time
For one focused interview loop: every mock and the AI Mentor while it lasts.
Best value
Annual
Save 75 %$12 / month
$147 billed once a year
A year for the price of three months. For prep that runs over months.
Lifetime
$441 once
one payment, about 3 years of annual
Every mock and the AI Mentor, and every question added later, for good.
Payments by Stripe. Cancel subscriptions any time from the billing portal; 7-day refund on a first purchase, under the refund policy.
What is a walkthrough?
A recorded system design interview between an interviewer and a strong candidate, played one turn at a time. You press Next, the candidate answers, and the whiteboard fills in as they talk: requirements, numbers, the API, then a design that gets fixed one bottleneck at a time, then the deep dives. At the key decisions it pauses and asks what you would say before showing what the candidate said.
What is the AI Mentor?
A chat beside the interview, never part of it. Ask it anything about what you have seen so far (why this database, what a term means, whether your own idea would work) and it answers with the interview and the whiteboard in front of it. Each turn also comes with suggested questions whose answers are written in advance; those are free for everyone. Live questions are part of Pro.
What is the mock interview?
You are the candidate. An AI interviewer runs the round the way a real one does: it states the problem, asks one thing at a time, probes what you said and what you drew on the whiteboard, and ends with a debrief (a score out of 100, what was strong, what was missing, and a note on each stage with a link to the same stage of the walkthrough).
Timed or with the AI Mentor?
Your choice at the start. Timed is interview conditions: a 45-minute clock the interviewer owns, and the AI Mentor is off. With the AI Mentor there is no clock and the AI Mentor sits beside you; it gives hints rather than answers unless you ask it outright, and the debrief notes that you had the AI Mentor.
What is free?
The Design a URL Shortener walkthrough, with no account, including the pauses and the suggested questions. Sign in with Google and every other walkthrough is free too. On a mock interview, anyone can hear the interviewer open the round. Which companies ask each question is free too. Pro adds live questions to the AI Mentor and answering the interviewer.
How is this different from reading a system design book or blog?
Reading gives you the final answer. A walkthrough shows you how a strong candidate gets there out loud (the order, the pacing, the numbers, and the design being fixed one problem at a time) and lets you stop and ask why. The mock then makes you produce it yourself, under pressure, with someone following up on the weak part.
What does Pro include?
Live questions to the AI Mentor, including comparing your answers with the candidate’s; AI mock interviews, timed or with the AI Mentor; and the company-style mock interviews. All three plans include the same thing; they differ only in how you pay.
Does the subscription renew automatically?
Monthly and annual plans renew until you cancel. Cancel from the billing portal at any time and you keep access until the end of the period you paid for. Lifetime is a single payment with no renewal.
Can I upgrade from monthly to annual or lifetime?
Yes. Open the billing portal from the pricing page and switch plans; Stripe prorates the unused part of the current period.
Is there a refund policy?
Yes, on a first purchase. Email us within 7 days of it, having started fewer than 3 mock interviews, and it is refunded in full. Renewals are not refundable, so cancel before the renewal date if you are done. The Terms have the details.
Do I need an account?
Not for the free walkthrough, and not to hear the AI interviewer open a round. Signing in with Google opens every walkthrough, and is needed to buy Pro.
Where do the "asked at" company logos come from?
Public interview guides and candidate reports. They indicate where a question has been reported, not a verified dataset, and a company name never implies endorsement.
What level is this for?
Senior and staff loops at large tech companies, where the interview is 45 to 60 minutes, expects numbers, and spends most of its time on deep dives. Mid-level candidates use it too; the walkthroughs explain every term as they go, and the AI Mentor will explain anything they do not.
Walk through one tonight. Sit one tomorrow.
One walkthrough is free with no account, and signing in opens all of them. Pause at any turn and pick it up later.