Teach

Why this section exists

I have run close to a thousand technical interviews over the last ten years, first at Amazon and now at Oracle Cloud Infrastructure. Coding rounds, system design rounds, the deep dive where we pull one thread of your past project until it snaps or holds. I still enjoy it. Not for the power trip. For the moment, about twenty minutes in, when a nervous candidate stops performing and starts thinking out loud. That is the whole job. Everything before that is theatre.

This section is for the people on the other side of that table. It is separate from the blog on purpose. The blog is where I write about GPUs, data centres, and whatever else is on my mind. This is a workbook.

It also keeps my own mind fresh. Writing a thing down is the only way I have ever remembered it. If I cannot explain merge intervals to my kids in two minutes with a drawing, I do not really know merge intervals. So every page here is built the same way: the question, a picture that moves, one trick, the steps, a template you can memorise, and a small box of what to remember. Short enough to read at a bus stop.

A note on the dates

I started these notes in 2016, on paper, as a way to remember what I kept asking and what kept tripping people up. The oldest pages here are those notes typed up, and the dates on them are honest. The GPU and AI track, and the short “In GPU infrastructure” paragraph you will now find in every coding and systems lesson, came much later, once my day job became a fleet of GPUs rather than a fleet of web servers.

What surprised me when I went back through them is how little I had to change. Two Sum is still Two Sum when the things you are pairing are half-empty GPU hosts. A sliding window is still a sliding window when the stream is NCCL timings instead of stock prices. The hardware got a lot more expensive and the fundamentals did not move at all. That is the point of the GPU paragraphs: not to make the old problems sound modern, but to show you that the shape you learned for the interview is the shape you will use on the floor.

What is here

Three tracks. Each lesson is one problem, one page.

• Coding. The classics. Not because interviewers are lazy, but because a classic tells me in fifteen minutes whether you can hold a data structure in your head and talk while you type.
• Systems. The design round. Load balancers, rate limiters, pipelines, schedulers. Boxes and arrows, but the interesting part is always the question I ask after you draw them: what breaks first.
• GPU and AI. The round that did not exist ten years ago. What a GPU actually is, how a thousand of them talk to each other, and how you know a brand new cluster is healthy before a customer finds out it is not.

How AI changed the interview

Everything and nothing. The tools got better. You can generate a working LRU cache in four seconds. So the question I ask moved. I no longer care whether you can recall the code. I care whether you can read what came back and tell me what is wrong with it. Whether you know the invariant well enough to test it. Whether, when the tool confidently produces the wrong thing, you notice.

The bar on fundamentals went up, not down. A tool amplifies what you already know. If you know nothing, it amplifies that too. So the lessons here are about the fundamentals. Each one ends with a line on how the question changes when there is an AI on the table, because there will be one.

How to use this

1. Read the question. Close the page. Try it on paper for ten minutes.
2. Come back and read the trick. If it matches yours, good. If it does not, that is the lesson.
3. Read the “remember this” box out loud. Yes, out loud. It works.
4. Come back in a week and do it again without the page.

The plan for the design round

Before the systems or GPU lessons, read how to approach a system design question. It is the plan every design lesson here follows.

Books

I keep a short list of the books I actually recommend to senior and staff candidates, with links to buy them. It is on the book list page.

The small print

None of the questions here come from a live interview loop at any company I have worked for. They are classics you will find in every book on the list above, retold in my words with my drawings. Everything here is my personal view and does not represent my employer.