Prompt writing and prompt evaluation jobs
Overview · 1 week ago
Prompt writing and prompt evaluation jobs pay you to write hard, realistic questions for AI models and to check whether a question is good enough to test one. What a task involves and what 36 listings on this site pay.
What prompt writing and prompt evaluation jobs are
Most evaluation work hands you a model's answer to judge. Prompt writing and prompt evaluation jobs work on the other end: you write the questions, and you judge whether a question is good enough to test a model at all. A strong answer to a weak prompt teaches nothing, so platforms pay for prompts that are realistic, hard and unambiguous. For the wider map of this work, see what AI training work actually is.
What one task looks like
The clearest description on the board is micro1's Evaluation Specialist (Recent Grad) listing. Each task is an original question and answer pair:
- Draft a multi-step question that needs synthesis across sources, not one search result.
- Research the answer and confirm it against several independent sources.
- Run the question against AI models. If a model gets it right, make it harder; if it proves ambiguous, tighten it.
- Document the answer with citations and reasoning, and fold in reviewer feedback.
Benchmark projects add structure. Mercor's Applied Biology Benchmark Specialist asks for one correct answer plus nine plausible but subtly wrong options, a step-by-step solution and one to five academic references, with each question rated for difficulty. That listing also shows the evaluation side: some contributors are assigned to verification instead, checking other people's questions for accuracy and rigour and documenting every edit.
In professional fields the prompts are scenarios. A micro1 AI Domain Expert listing, since closed, asked for prompts built on realistic situations that push the model's reasoning rather than trivia it can recall, followed by a written account of what the model should have done.
Who hires for it
Of the 699 listings on this site on 29 September 2026, 36 name prompt writing, prompt engineering or question writing as a skill or in the title. Mercor posted 26 and micro1 10.
More than half (22) belong to Mercor's bilingual AI safety family, where native speakers write prompts in their own language on sensitive subjects and classify how a model should handle them. The rest are benchmark and domain expert roles.
What it pays
All 36 quote an hourly band, from $14 to $200 an hour. The median band runs from $48 to $60.
- Question writing on any subject: micro1's Evaluation Specialist at $20–60, Mercor's research question role for Apple Silicon MacBook owners at $50–60.
- Bilingual safety prompts: $18–81 depending on language and whether the role asks for a STEM PhD.
- Science benchmarks: $60–75 for applied biology, $61–77 for applied physics.
- Professional domains: micro1's AI Domain Expert at $140–200, and its finance, consulting, data science and software domain experts at $100–200.
The low end, $14–36, is micro1's Senior AI Trainer, where writing edge-case prompts is half of a generalist grading job.
Worth knowing
Per-task pay changes the arithmetic here more than anywhere. The micro1 recent-grad listing pays per task meeting the specification, and a question the model answers easily, or one that fails review, is time spent without pay. Expect early earnings near the floor while you learn what passes. Our payout guide explains the model.
People who have written exam questions, set quizzes or fact-checked for a living have a head start. Browse current roles under AI Safety & Evaluation and Writing & Education.
Questions
- What does a prompt writing job involve?
- You draft an original question that a model should find hard, research and verify the answer, test the question against AI models, tighten it if a model gets it right or if it is ambiguous, and document the answer with sources and reasoning.
- What is prompt evaluation?
- Checking other people's prompts and questions for accuracy, rigour and clarity, editing them where needed and documenting every change. Mercor's benchmark projects assign some contributors to this verification work instead of authoring.
- How much do prompt writing jobs pay?
- On this site on 29 September 2026, 36 listings named prompt or question writing, all quoting hourly bands from $14 to $200. The median band ran from $48 to $60. Professional domain roles paid the most.
Platforms covered here
Put this into practice
Every listing shows its pay and who it is open to.