The Interview That Ships to Production: replacing whiteboards with pull requests

3 days ago

AngelList replaced traditional whiteboard interviews with a practical take-home coding exercise and pull request-based follow-up.
The interview process involves implementing a venture fund distribution waterfall in JavaScript, requiring algorithmic thinking and precision.
Candidates use an AI assistant (Claude) intentionally, testing their ability to evaluate and verify AI output rather than avoid using it.
Candidates who pass the take-home are asked to make a pull request on the actual codebase, improving a real aspect of the testing platform.
Several examples of candidate improvements include UI enhancements, bug fixes, and AI calibration tweaks that were merged into production.
The process self-improves as each candidate's contributions enhance the platform for future candidates.
This approach aligns with AngelList's work culture: small team, high impact, and agency to identify and ship meaningful changes.
The interview signals the company's values, like skepticism, critical thinking, and leveraging AI responsibly.

Hasty Briefsbeta