Merge
AI-native code review assessments for engineering hiring
Last verified:
What is Merge?
Merge is an AI-powered code review assessment platform for technical hiring that lets candidates review pull requests in a realistic codebase while an AI agent responds to their comments in real-time, simulating actual code review iterations. It provides hiring signals based on real engineering work—how candidates identify bugs, refactors, security risks, and iterate on code quality—rather than algorithm quizzes.
Merge pricing
Pricing model: Freemium
Merge pros
- Realistic assessment: Candidates perform actual code review work similar to the job, not algorithmic puzzles
- Custom calibration: Adjustable difficulty levels (intern to principal), specialization (frontend/backend/infrastructure/etc.), and language restrictions for role-specific hiring
- Token efficiency tracking: Shows how efficiently candidates use AI and provides cost visibility and iteration metrics
- Detailed reporting: Scorecards connect candidate reasoning to code quality, risk detection, and actionable hiring recommendations
Merge cons
- Pricing not visible on site; requires demo booking for quotes
- Time-intensive process requiring multiple iterations, which may extend hiring timelines
- No visible information on integration capabilities with existing ATS or hiring platforms
Frequently asked questions about Merge
How does Merge assess engineering ability?
Candidates review a realistic pull request and comment on correctness bugs, refactors, and security risks. An AI agent then addresses their feedback with a PR revision, and the loop continues until time expires or the candidate approves the code.
What can be customized in assessments?
Each assessment can be calibrated by difficulty level (intern to principal), specialization (frontend, backend, infrastructure, etc.), and supported programming languages.
How does Merge track candidate efficiency?
Merge shows exactly how efficiently candidates work with AI, including token usage, estimated costs, and the number of PR revisions required.