Inside the Tech Giants

About this course

Big tech companies publish more about their systems than most people read: papers, engineering blogs, talks and postmortems. This course reads them carefully. Each lesson takes one giant apart, states the problem it had to solve, follows a request through the system, explains the data layer the way a database administrator would, and tells the story of a real outage. Every fact carries its year, and every lesson ends with a box that separates what is published from what is a reasonable guess.

You then rebuild the heart of each system yourself, in Postgres, Redis, TypeScript and ffmpeg: a sharded inverted index with hedged requests, a cache with memcache-style leases, Instagram's 64-bit ids, a WhatsApp-style inbox with ticks, a nearest-driver ring search, an oversell-proof stock reservation and an adaptive-bitrate player rule. Auto-checked JavaScript and SQL exercises let you practise the core logic in your browser.

Every lesson ends at Mangrove Market, the fictional grocery chain with 320 stores, 1,800 partner shops and 600 riders used across Kolom, with what to borrow now, at 10 times and at 100 times its size, and what never to copy. The last lesson pulls the patterns together into a scale ladder and a review template, and the capstone asks you to use them on a real design.

What you will be able to do

• Explain how each of nine large systems solves its hardest problem, citing public sources and their year
• Rebuild a giant's core mechanism at small scale: an inverted index, a lease cache, a 64-bit id, an inbox with receipts, a ring search, a conditional stock reservation and an adaptive-bitrate rule
• Separate published facts from guesses when you read an engineering blog, paper or postmortem
• Place a system on the scale ladder from its own numbers, and say what changes at 10 times and 100 times
• Decide which giant's ideas to borrow now, which to leave for later and which never to copy
• Answer classic interview questions: design search, a feed, a chat app, ride-hailing, a checkout and video streaming

Syllabus

10 stops, about 6.5 h in total. Each stop fills in when you complete the lesson.

1. 1

   Google Search: crawling, indexing and answering in a fraction of a second

   In March 2025 Google said it sees more than 5 trillion searches a year, which is on the order of 150,000 every second, each answered from an index of hundreds of billions of pages. This lesson takes the system apart: how pages are fetched, how the index is built and split across thousands of machines, and how one query is answered by many servers at once without waiting for the slowest one.

   40 minAdvancedEpisode2 exercises

   Inverted indexes and searchTail latencySharding and IDsArchitecture judgement

2. 2

   Facebook: the social graph, TAO and memcache at planet scale

   In June 2026 Meta reported 3.6 billion people using at least one of its apps every day, and its graph store, TAO, was serving over ten billion reads a second back in 2021. This lesson shows how that is built from parts you already know: MySQL, a look-aside cache, sharding and replication, plus a few clever fixes for the problems that only appear when a cache is that big.

   40 minAdvancedEpisode2 exercises

   Cache invalidationSharding and IDsArchitecture judgement

3. 3

   Instagram: from 3 engineers on Postgres to a billion users

   In 2011 Instagram had 14 million users and 3 engineers; in September 2025 Meta said it had 3 billion monthly users. In between, its core stayed surprisingly plain: Django, PostgreSQL split into thousands of logical shards, Redis and memcached, and a 64-bit ID made inside Postgres. This is the giant whose early stack looks most like your team's, so it is the one to study most closely.

   40 minAdvancedEpisode2 exercises

   Sharding and IDsFeed fan-outArchitecture judgement

4. 4

   WhatsApp: billions of messages with a tiny team

   In 2020 WhatsApp was delivering roughly 100 billion messages a day, and in 2025 Meta said more than 3 billion people use it every month. In 2014 its server side was run by about ten Erlang engineers. This lesson shows the few simple ideas that made that possible, and builds a small version you can run.

   40 minAdvancedEpisode2 exercises

   Reliable messagingEnd-to-end encryption basicsArchitecture judgement

5. 5

   Uber: matching riders and drivers in real time

   In the third quarter of 2025 Uber completed about 3.5 billion trips, roughly 38 million a day, for 189 million monthly users. Every one began with the same few seconds: find nearby drivers, estimate how long each would take, and offer the trip to one. This lesson takes those seconds apart and builds the "nearby drivers" part with hexagons.

   40 minAdvancedEpisode2 exercises

   Geospatial matchingLive state in memoryArchitecture judgement

6. 6

   Amazon: the store that never closes

   On Prime Day 2025, DynamoDB tables behind Amazon's systems peaked at 151 million requests a second. Amazon is the giant closest to your own business: a retailer with carts, stock, orders and warehouses. This lesson shows how its cart, inventory and order systems are built from public sources, and builds the stock reservation you would actually use.

   40 minAdvancedEpisode2 exercises

   Stock reservationIdempotencyKey design for key-value storesArchitecture judgement

7. 7

   Netflix: streaming to every screen

   On 15 November 2024, a single boxing match on Netflix reached 65 million streams at the same moment. In 2025 Netflix said its own delivery network has more than 18,000 servers in over 6,000 locations. This lesson shows the main trick behind that: a small "brain" in the cloud decides, and a huge, simple network of boxes inside internet providers carries the video. Then you build the same ladder of video qualities with ffmpeg.

   40 minAdvancedEpisode1 exercise

   Video deliveryAdaptive bitrateControl and data planesArchitecture judgement

8. 8

   ChatGPT: serving a large language model to hundreds of millions

   In October 2025, OpenAI's CEO said more than 800 million people use ChatGPT every week, and its API processes over 6 billion tokens a minute. Most of ChatGPT's internal design is private, but OpenAI has published real pieces: its PostgreSQL setup, its Kubernetes clusters, its outages, and how its engineers think about GPUs. This lesson puts those pieces together with how LLM serving works in general, and then you build a streaming chat proxy you can put in front of any model.

   40 minAdvancedEpisode2 exercises

   LLM servingStreaming responsesRate limitingCapacity planning

9. 9

   Claude: how Anthropic serves Claude at scale

   In 2025 Anthropic said it serves Claude to millions of users on three kinds of chips, from three cloud partners, and to more than 300,000 business customers. Anthropic publishes little about its internal systems, but its API documentation and one detailed postmortem show a lot about how a model service behaves from the outside, and what went wrong inside. This lesson reads those sources carefully, separates them from general knowledge, and ends with a resilient client you can use with any LLM API.

   40 minAdvancedEpisode1 exercise

   Prompt cachingResilient API clientsLLM servingAI evals

10. 10

    What the giants share, and what not to copy

    Nine companies serving between hundreds of millions and billions of people use a surprisingly short list of ideas, and almost every idea is one you met in the earlier System Design courses. In 2026, OpenAI described the PostgreSQL behind ChatGPT running on a single primary with read replicas. This lesson pulls the patterns together, shows at what size each one starts to pay off, teaches you to read an engineering blog critically, and ends with a review template you can use for a real system, filled in for Mangrove Market.

    40 minAdvancedEpisode1 exercise

    Architecture judgementCapacity estimationReading engineering blogsDesign reviews

11. Capstone: Make Mangrove's Now delivery ready for the festival rush

    A realistic project with a rubric and a model answer.

12. Final exam and certificate

    10 questions drawn at random, 70% to pass.

Capstone project

Check what you know

Questions