---
schema_version: "newruntime-agent-readable-v0.1"
type: "post"
slug: "coinbase-ai-era-interviews"
title: "Coinbase Rebuilds Interviews for Engineers with AI"
description: "Coinbase describes interviews that test not the ability to code without help, but the ability to direct AI, review output, and make engineering decisions in a new work loop."
status: "published"
published_at: "2026-07-21"
topics: ["engineering-hiring","ai-adoption","developer-productivity"]
source_urls: ["https://www.coinbase.com/en-gb/blog/interviewing-engineers-in-the-ai-era-lessons-from-a-year-of-rebuilding"]
routes: {"html":"https://newruntime.com/posts/coinbase-ai-era-interviews/","markdown":"https://newruntime.com/posts/coinbase-ai-era-interviews.md","json":"https://newruntime.com/posts/coinbase-ai-era-interviews.json"}
source_format: "markdown"
---

# Coinbase Rebuilds Interviews for Engineers with AI

Coinbase describes a year of rebuilding engineering interviews for an era where AI is already part of work. That is a useful signal: companies are starting to test not only raw coding, but how an engineer manages AI output.

In the old model, the interview tries to isolate the human from tools. In the new work reality, that isolation is artificial. A production engineer will read unfamiliar code, give an agent context, catch errors, make architecture decisions, and own the result.

## What to watch

The evaluation shifts toward:

- how the candidate frames work for AI;
- how generated code is checked;
- when output is accepted and when it is rewritten;
- how tradeoffs are explained;
- how quality, security, and system style are preserved.

## New Runtime Read

An AI-native interview should test engineering judgment around the agent. If a company still bans AI in interviews, it may be testing a skill that is less relevant to the next morning's work.