RLHF and AI Alignment Jobs

Read this page with the numbers in front of you. As of September 2026 the RLHF and alignment hub holds 36 live listings, none posted in the past seven days, and 35 of them are the same freelance project repeated across languages: Language Alignment and Resource Partner roles for Portuguese, Icelandic, Khmer, Korean, Marathi, Lithuanian and others, posted through one agency account. The remaining listing is a machine learning engineer role at OKX.

Those language alignment projects are human feedback work in the literal sense: native speakers rating, correcting and writing model outputs so that a system can be trained on preference data. They are categorized here as AI annotation and data labeling, they are almost all remote (89 percent), tagged mid level, and they state no pay in US dollars. They are not research positions, and nothing on this hub currently describes reward modeling or policy optimization as an engineering job.

If you are looking for the research and engineering side of RLHF, the roles exist on TixelJobs but are filed elsewhere. /reinforcement-learning-jobs collects RL titled positions, /ai-safety-jobs holds alignment and safety teams, /research-scientist-jobs and /llm-jobs carry the post-training and evaluation work at model labs, and a keyword search for alignment, preference or post-training will surface roles whose titles say something else.

If freelance language work is what you want, this hub is a decent index of it: the listings span low resource and high resource languages alike, and the country tags (one each for India, the US, Canada and Singapore) confirm that location is mostly irrelevant. Check /freelance-ai-jobs for the wider set of similar projects, since the same agency posts several hundred there.