Interview Formats

AI-enabled coding interviews come in two flavors, and knowing which one you're walking into matters a lot. The tools available, the types of problems you'll face, and even the strategies that work best are all different depending on the format.

Often you can tell from the invite itself. A link to a browser-based environment they've set up points to structured. Being told to use your own machine and share your screen points to open-ended. If you're not sure, ask your recruiter — and if they don't know, push until you get an answer.

Structured (CoderPad, HackerRank)

The company provides the coding environment and the AI assistant. You open a browser, log in, and everything is already there: the codebase, the AI chat panel, and the problem statement loaded before you write a line.

These interviews follow a defined problem arc — typically three phases:

Knowing that structure early lets you pace yourself. You know roughly what's coming and can avoid burning all your time on phase one.

Who runs it

• Meta is the most well-known example. Their CoderPad setup has become the reference point candidates compare everything else to — a selection of AI models (GPT-4o, GPT-5, Claude Sonnet/Haiku, Gemini, Llama).
• LinkedIn also uses CoderPad, with their own model selection and the same multi-phase structure.
• Uber runs on HackerRank with a pre-set repository, where you're expected to navigate existing code before making changes.

Not every structured round is build-and-optimize. A few companies hand you a pull request instead and ask you to use the AI to work through its description and diff, and decide whether you'd approve it. Same skills, different shape.

Both Meta and LinkedIn offer practice sessions on their platform before the real interview. Always take advantage of these. The CoderPad AI interface has quirks you won't encounter in your normal IDE, and spending 30 minutes getting comfortable with it beforehand is time well spent.

Open-ended (Bring Your Own Environment)

You use your own IDE, your own AI tools, and share your screen. This format is more common at companies that want to see how you actually work day-to-day — they're not interested in whether you can adapt to a constrained platform; they want to know what your real workflow looks like.

• Shopify is the canonical example: they tell you to use whatever you'd normally use at work.
• Canva mandates AI usage and scores it explicitly.
• Rippling has you build features incrementally against test cases provided by the interviewer.
• Adobe varies by team, from structured DSA to open repo work.

The problems here are greenfield. You might get an empty repository and a problem statement, or just a README describing what to build. You start from nothing and build a working application. The emphasis shifts from "can you navigate existing code" to "can you design and build something from scratch with AI as your accelerator."

Open-ended problems tend to pull in low-level design skills: designing software that is both functional and extensible, even when most of the implementation is handled by the AI.

What to ask your recruiter

Before the interview, get clear answers to these questions:

The last three are newer, and they matter more than they sound.

Tokens. In a bring-your-own-tools round you may be expected to burn your own API credits to interview. Some companies provide an account; plenty haven't thought about it. Ask, and let them answer.

Your own config. If you normally work with a rules file and a set of saved skills, whether you can bring them changes your preparation completely. Some interviewers consider that setup part of your craft and want to see it; others want a clean environment so they're grading you rather than your dotfiles. Nobody will volunteer this.

What they're grading. Ask directly, in the first few minutes if you didn't get an answer beforehand.