CSE Roadmap

Common tier · about 6 hours · 7 parts

Part 2 of 7 · 2 min read

Method

A fixed routine for each prompt.

Think of practice scales for a musician. The same routine, run on many pieces, builds speed and calm.

The routine (45 minutes per prompt).

  1. Minutes 0 to 5: scope. Write 4 questions and your assumed answers. Write the main user actions.
  2. Minutes 5 to 10: numbers. Users, requests a second, data a day, data in total, peak. Write the formula for each.
  3. Minutes 10 to 20: first design. Draw the simplest design that works for one user. Number the steps of the main flow.
  4. Minutes 20 to 35: deep dive. Find the bottleneck and fix it. Choose the one or two hardest parts and design them.
  5. Minutes 35 to 42: failures and trade-offs. What breaks? What do you give up? What would you monitor?
  6. Minutes 42 to 45: summary. State the design in six sentences.

Check afterwards. Compare with the "core idea" and the checklist below. Mark each point as found, partly found or missed. Write what you missed on a list, and revisit the matching earlier level.

Difficulty marks. E (easy, a few parts), M (medium, one deep part), H (hard, several deep parts).

Prompts are not on the same footing as the full case studies. The "core idea" line is only a signpost. An answer can differ and still be good, if it is consistent, has numbers and names trade-offs.

Core idea

The core idea of a design is the one decision that makes the problem tractable, from which most other choices follow.

Example: for a URL shortener, the core idea is a short unique id from a counter or a hash, stored in a key-value table that is read far more than written.

Check yourself: 1. What happens in minutes 0 to 5? 2. Why write the formula for each number? 3. Is a different answer from the core idea always wrong?