Scoring an answer without a keyword list
Keyword matching is the oldest trick in automated screening, and it's popular for a reasonable reason: it's cheap to build, easy to explain, and fast to run. It's also trivially easy to game — pad a resume or an answer with the right nouns, and a keyword scorer can't tell the difference between someone who used a technology and someone who mentioned it. Worse, it actively penalizes a strong answer that happens to use different words than the ones on the scorer's list.
The Interview Evaluator doesn't work that way. It weighs an answer against the actual competencies a role calls for — substance, structure, and relevance — closer to how a good hiring manager listens than to how a search engine ranks documents.
Substance over vocabulary
A candidate who says "I noticed our deploys were failing intermittently, traced it to a race condition in the queue consumer, and fixed it by adding idempotency keys" should score well on a reliability-engineering question whether or not they ever say the word "reliability." A candidate who says all the right buzzwords with no concrete detail behind them shouldn't. The evaluator is built to tell those two answers apart, because the difference between them is the entire point of asking the question.
One standard, every format
RiseYou's interviews aren't all the same shape. A screening call is a five-minute structured conversation. The AI Interview Round can include a technical exercise with real submitted code, a system-design walkthrough, or a mixed assessment block. A keyword-based scorer would need a different, brittle ruleset for each format. The Evaluator applies the same competency-weighted standard across all of them — a candidate's code submission and their spoken answer to a behavioral question are judged by the same underlying discipline, not two disconnected systems bolted together.
Show the reasoning, not just the number
A bare score is a black box, and black boxes don't build trust with the hiring teams who have to stand behind a decision. The Evaluator surfaces why it scored an answer the way it did — which parts of the competency were demonstrated, which weren't — so a hiring manager reviewing the result isn't just trusting a number, they're checking its work.
The evaluator weighs substance, structure, and relevance the way a good hiring manager would — and shows its reasoning, not just a number.
None of this makes the evaluator immune to being wrong about a specific answer — no evaluator, human or otherwise, is. It's built to be wrong for the right reasons, and to make those reasons visible when it is.