Review: interview feedback hire signal + evidence

The job

Jev judge eval (1 choice + 1 score) over a fixed state : runnable via experimental_evaluate.

The prompt

Copy this into Muse and replace anything in [square brackets] with your own details.

// Review: interview feedback hire signal + evidence — Jev judge eval (typesafe-ai/jev).
// A panel writes feedback on a backend candidate; the judge reads the signal and checks evidence.
// Model: typesafe-ai/jev via the Vercel AI Gateway ($0.042/1M input tokens).
// Auth: set AI_GATEWAY_API_KEY (vck_…) in the environment — never paste the key here.
// For zero-data-retention / no-training workloads, add
// providerOptions: { gateway: { … } } to the evaluate call below.

import { experimental_evaluate } from 'ai';

const result = await experimental_evaluate({
  model: 'typesafe-ai/jev',
  state: "Feedback: Strong hire. System design was excellent — she drove the sharding discussion, named the hot-partition risk unprompted, and sketched a clean migration. Coding: correct optimal solution in 25 min with clear tests. Concern: none material; ramp-up on our queue infra expected.",
  questions: {
    "signal": {
      "type": "choice",
      "instructions": "Classify the overall hire signal of this feedback.",
      "criteria": {
        "strong_hire": "Enthusiastic yes with concrete standout evidence.",
        "hire": "Yes, with solid but not exceptional evidence.",
        "no_hire": "No, with concrete gaps or red flags.",
        "mixed": "Contradictory or insufficient evidence either way."
      }
    },
    "evidence_quality": {
      "type": "score",
      "instructions": "Score how evidence-backed this feedback is. Higher is better.",
      "criteria": [
        "Verdict with no examples.",
        "Verdict with vague praise (great, smart).",
        "Verdict with one concrete example.",
        "Verdict with multiple specific, observed examples plus a named non-concern."
      ]
    }
  },
});

console.log(JSON.stringify(result, null, 2));