Free course · 39 lessons · 20 mock interviews

System Design: zero to interview-ready

From what happens when you open an app to designing WhatsApp, Uber and ChatGPT. Every lesson has its video, the whiteboards, the notes and a quick check. Each week ends with a revision and a full mock interview, written out turn by turn with the interviewer's feedback.

26 videos outNew videos at 6 pm

Start here

What the course covers, and how to use it, in one short video.

Your progress0 of 67 done · saved on this device

Week 1: Foundations

  1. 1What happens when you open an appClient and server; DNS turning a name into an IP address; the TCP and TLS handshakes; one HTTP request and its response; what each hop costs in millisecondsVideo
  2. 2The numbers every engineer should knowLatency numbers (memory, SSD, a round trip in one data centre, one across a continent); latency versus throughput; p50 versus p99; availability nines (99.9 percent is about 8.8 hours down a year); then back-of-the-envelope estimation: daily users to requests a second (a day is about 100,000 seconds), peak as two to three times the average, storage a yearVideo
  3. 3APIs and data models: the contract of a systemREST resources, verbs and status codes; cursor pagination; idempotency keys for safe retries; when gRPC, GraphQL or WebSockets fit; then data modelling: entities, primary keys and relationships, and SQL tables versus documents versus key-value in one pictureVideo
  4. 4The 45-minute system design interviewThe framework every answer follows: requirements (functional and non-functional), core entities, the API, the high-level design, then deep dives; how long each step gets; what interviewers score at mid, senior and staff level; the common mistakes (boxes before requirements, no numbers, no trade-offs)Video
  5. RWeek 1 revisionThe whole week on a few boards, with quick checksVideo
  6. M1Design a URL shortener (Bitly)Mock interview 1: Design a URL shortener like Bitly.Video

Week 2: Scaling

  1. 5Scaling from one server to manyVertical versus horizontal scaling and where each stops; stateless app servers; moving sessions to a shared store; why the database becomes the next bottleneckVideo
  2. 6Load balancersLayer 4 versus layer 7; round robin, least connections and hashing; health checks; removing the single point of failure with an active and a standby load balancerVideo
  3. 7Caching: making reads fastWhere caches sit (browser, CDN, application, database); cache-aside versus write-through versus write-back; TTLs; eviction with LRU; what the hit rate does to latencyVideo
  4. 8Caching at scale: invalidation, hot keys and stampedesA shared cache like Redis; invalidation strategies and their failure modes; hot keys (replicas, a small local cache); the thundering herd and request coalescing; warming a cacheVideo
  5. 9CDNs and object storageWhy files belong in object storage (S3) and not the database; presigned URLs for uploads; how a CDN caches at the edge; cache-control headers; image variantsVideo
  6. RWeek 2 revisionThe whole week on a few boards, with quick checksVideo
  7. M2Design Instagram: posting and viewing photosMock interview 2: Design Instagram, posting and viewing photos.Video

Week 3: Data

  1. 10Database indexes: how lookups stay fastA full scan versus an index; a B-tree in one picture; composite indexes and the leftmost prefix; what every index costs on writes; LSM trees for write-heavy storesVideo
  2. 11Replication: copies of your dataLeader and followers; synchronous versus asynchronous; read replicas and replica lag (read-your-own-writes); failover and promoting a follower; multi-leader in one lineVideo
  3. 12Sharding: splitting data across machinesWhen one machine can no longer hold the data; partition keys; range versus hash partitioning; hot shards; queries that cross shards; reshardingVideo
  4. 13Consistent hashingWhy hash modulo N breaks when N changes; the ring; virtual nodes for balance; where it is used (cache clusters, Dynamo, Cassandra)Video
  5. 14CAP, consistency and quorumsStrong versus eventual consistency; what CAP really says during a network partition; quorum reads and writes (N, R and W); choosing per feature (a payment versus a like)Video
  6. RWeek 3 revisionThe whole week on a few boards, with quick checksVideo
  7. M3Design a key-value store like DynamoDBMock interview 3: Design a key-value store like DynamoDB.Video

Week 4: Async

  1. 15Message queuesWhy to put a queue between services; producers and consumers; at-least-once delivery, acknowledgements and visibility timeouts; retries with backoff; dead-letter queuesVideo
  2. 16Kafka and event streamingA log versus a queue; topics and partitions; consumer groups; ordering within a partition; retention and replaying eventsVideo
  3. 17Idempotency and exactly-once in practiceWhy duplicates always happen; idempotency keys; deduplication tables; the outbox pattern; exactly-once as at-least-once plus idempotent processingVideo
  4. 18Real-time updates: polling, SSE and WebSocketsShort polling, long polling, server-sent events and WebSockets; connection servers; fanning updates out with pub/subVideo
  5. 19Rate limitingWhy to limit; token bucket versus fixed and sliding windows; where to enforce it (the gateway); shared counters in Redis; answering 429 with Retry-AfterVideo
  6. RWeek 4 revisionThe whole week on a few boards, with quick checksVideo
  7. M4Design WhatsAppMock interview 4: Design a chat app like WhatsApp.Video soon

Week 5: Architecture

  1. 20Microservices, gateways and service discoveryStart with a monolith and when to split it; what an API gateway does; service discovery; synchronous versus asynchronous calls between servicesVideo soon
  2. 21When things fail: timeouts, retries and circuit breakersFailure is normal; timeouts everywhere; retries with exponential backoff and jitter; circuit breakers; bulkheads; degrading gracefullyVideo soon
  3. 22Distributed transactions: two-phase commit and sagasWhy one transaction across services is hard; two-phase commit and why it blocks; sagas with compensating steps, orchestrated or choreographedVideo soon
  4. 23Unique IDs at scaleWhy auto-increment stops working; UUIDs; Snowflake IDs (time, machine and sequence); ticket servers; IDs that sort by timeVideo soon
  5. 24Observability: logs, metrics and tracesThe three pillars; RED and USE metrics; distributed tracing with a trace ID; SLIs, SLOs and error budgets; alerting on symptoms, not causesVideo soon
  6. RWeek 5 revisionThe whole week on a few boards, with quick checksVideo soon
  7. M5Design a payment systemMock interview 5: Design the payment system for a food delivery app.Video soon

Week 6: Patterns

  1. 25News feeds: fan-out on write or on readPush versus pull; the celebrity problem and the hybrid answer; caching feeds; ranking in one line; cursor paginationVideo soon
  2. 26Search: inverted indexes and typeaheadWhy LIKE '%pizza%' falls over; the inverted index; Elasticsearch shards and replicas; keeping the index in sync with change data capture; typeahead with prefixesVideo soon
  3. 27Geospatial: finding what's nearbyWhy latitude and longitude ranges are slow; geohash; quadtrees; H3 cells; caching nearby resultsVideo soon
  4. 28Contention: two people, one seatRace conditions; pessimistic locks; optimistic concurrency with version numbers; holds that expire; distributed locks and their limitsVideo soon
  5. 29Big files and long jobsChunked and resumable uploads; presigned multipart uploads; background workers and job queues; progress, retries and timeoutsVideo soon
  6. RWeek 6 revisionThe whole week on a few boards, with quick checksVideo soon
  7. M6Design UberMock interview 6: Design a ride-hailing app like Uber.Video soon

Week 7: Case studies

  1. 30Design YouTube: upload, transcode, streamUpload to object storage; a transcoding pipeline (queue and workers); adaptive bitrate streaming (HLS); delivering through a CDN; counting viewsVideo soon
  2. 31Design Dropbox: files in sync everywhereSplitting files into chunks; content hashes and deduplication; a metadata service; the sync protocol; conflicting edits; telling devices about changesVideo soon
  3. 32Design a web crawlerThe URL frontier; politeness with a queue per host; robots.txt; deduplicating URLs and content (Bloom filters); distributed workers; deciding when to crawl againVideo soon
  4. 33Design search autocompleteTop suggestions for every prefix; a trie with the top results cached at each node; building it offline from query logs; updating it; answering in under 100 millisecondsVideo soon
  5. 34Design a notification systemPush, SMS and email; a queue per channel; templates; user preferences; rate limits per user; retries and deduplication; third-party providersVideo soon
  6. RWeek 7 revisionThe whole week on a few boards, with quick checksVideo soon
  7. M7Design TicketmasterMock interview 7: Design a ticket booking system like Ticketmaster.Video soon

Week 8: Senior level

  1. 35Design Google Docs: editing together in real timeOperational transformation versus CRDTs; one document session per server; WebSockets; snapshots and versions; cursors and presenceVideo soon
  2. 36Design a distributed job schedulerCron at scale; storing schedules; workers claiming jobs with leases; at-least-once runs and idempotent jobs; priorities and retriesVideo soon
  3. 37Stream processing: an ad-click aggregatorClick events into Kafka; windowed counts with a stream processor like Flink; checkpoints for exactly-once; late events and watermarks; serving the countsVideo soon
  4. 38Multi-region systemsWhy go multi-region (latency, disasters, data residency); active-passive versus active-active; geo DNS; replicating across regions; resolving conflicts; failover drillsVideo soon
  5. 39Thinking like a senior engineerWhat separates a senior answer: depth where it matters, trade-offs said out loud, numbers as a habit, operations and cost; how to keep improving after the course; the whole course in one boardVideo soon
  6. RWeek 8 revisionThe whole week on a few boards, with quick checksVideo soon
  7. M8Design a food delivery appMock interview 8: Design a food delivery app like Swiggy or Zomato.Video soon

Final mock interviews

After the eight weeks: twelve more interviews, from Yelp to ChatGPT.

  1. M9Design Yelp: places near youMock interview 9: Design a local discovery app like Yelp.Video soon
  2. M10Design a distributed cacheMock interview 10: Design a distributed in-memory cache, like Redis Cluster.Video soon
  3. M11Design SpotifyMock interview 11: Design a music streaming app like Spotify.Video soon
  4. M12Design an online judge like LeetCodeMock interview 12: Design an online coding judge, like LeetCode.Video soon
  5. M13Design live comments for a streamMock interview 13: Design live comments for a live video stream.Video soon
  6. M14Design a train booking rushMock interview 14: Design a train booking rush, like India's Tatkal.Video soon
  7. M15Design live video streaming for 50 million viewersMock interview 15: Design live video streaming for fifty million viewers.Video soon
  8. M16Design Google Maps: routes and ETAsMock interview 16: Design Google Maps, its routes and arrival times.Video soon
  9. M17Design a stock exchangeMock interview 17: Design a stock exchange.Video soon
  10. M18Design a hotel booking systemMock interview 18: Design a hotel booking system, like Booking.com.Video soon
  11. M19Design a metrics and monitoring systemMock interview 19: Design a metrics and monitoring system.Video soon
  12. M20Design ChatGPT: serving a large language modelMock interview 20: Design the serving system for ChatGPT.Video soon

New videos go up on our YouTube channel at 6 pm.