Technical interviews, in context
Interview engineers
in real code.
Bring a repository to the interview. See how someone explores it, explains their thinking, and makes it better.
“What happens if we retry?”
“Let me write a test.”
- 01 Bring your exercise repository
- 02 Send the candidate a link
- 03 Talk through the work together
Watch how they
work with the codebase.
A checkout now retries failed payments. The tests pass. But something isn’t quite right.
Interviewer: “What happens if the payment fails?”
Candidate: “Could retrying reserve the stock twice?”
import { retry } from "./retry";
import { reserveStock } from "./stock";
import { chargeCard } from "./payments";
export async function checkout(
order: Order,
) {
return retry(async () => {
await reserveStock(order);
return chargeCard(order);
});
}
More time for the interview.
Less time getting set up.
A link, and they’re in.
The VS Code workspace they already know, with the terminal and project tests ready in the browser. No account, no local setup.
Their own copy of the code.
Every candidate starts from the same prepared exercise. Their changes stay in their workspace, away from your GitHub repository.
AI, when you want it.
Turn on autocomplete, a coding agent, both, or neither — per exercise. Either way, you can ask how the code works and how they know it’s right.
Make room for a better conversation
Bring your next
interview.
Prepare an exercise. Send a link.
See where the conversation goes.
Bring your own accounts.
Workspaces run on Modal. Bring your own AI provider for autocomplete and agents, or turn them off. You’ll need your own service accounts.
Prefer to run it yourself?
Run Chunin on your own infrastructure with Docker Compose. Keep the interviewer app on your private network and share candidate links separately.
Read the setup guide