What it means
Also called: reinforcement learning from human feedback.
A way of training a model on people's judgements. People compare or rate the model's answers; those judgements train a reward model, which is then used to train the model towards answers people prefer. Most rating and ranking work feeds into it.
Used in: AI evaluator and rater roles, and expert roles in most fields.
In open roles right now
72 of the open roles we list mention “RLHF” or a word for it in the title or description, on 11 platforms (as of 23 October 2026). Of those that state an hourly rate (26 distinct postings), the median is $70/hr, and the middle half pay $45 to $105.
Most of them on: Alignerr (35), Turing (9), SuperAnnotate (8), micro1 (7), Volga Partners (6), Prolific (2), and 5 more.
Domain Contributor - Evaluation Specialistmicro1·Location not stated·Part-time
$30–60/hr2 days ago
Domain Expert – PythonTuring·10 countries·Flexible hours
Pay set per project15 days ago
AI Safety Experts - English & FinnishWeekday·United States
$48–62/hr1 month ago
AWS Trainium / NKI Kernel ExpertAnyone AI·9 countries·Part-time
$65/hr1 month ago
Knowledge Expert — Human Communication & Context ElicitationVerita AI·10 countries·Full-time·PhD
$28 per task2 months ago
Related terms
EvaluationTesting what a model can do by having it answer a set of questions and scoring the answers, often by people working from a rubric.Fact-checkingChecking each claim in a model's answer against reliable sources and marking what is wrong, unsupported or out of date.Preference rankingComparing two or more model answers to the same prompt and saying which is better, often with a reason and a rating of how much better. ← RL environmentPrevious term, A to Z. Rubric →Next term, A to Z. The whole glossaryEvery term on one page, A to Z.Open →