RLHF Data Operations Associate
The RLHF Data Operations Associate role at Corpshore AI's Samarkand, Uzbekistan hub supports LLM alignment by comparing model outputs against a rubric to produce preference data, an onsite contract position with full paid training and no AI background required.
Samarkand is Corpshore AI's newest Central Asia seat expansion, standing up RLHF and preference-labeling capacity alongside the Tashkent studio. As an RLHF Data Operations Associate, you'll compare model outputs against structured rubrics, write short rationale notes, and keep the operational side of preference-data batches (routing, timestamps, batch handoffs) moving cleanly through Jwuma.
Key responsibilities
- Compare pairs of model responses against a rubric and record structured preference judgments
- Draft concise rationale notes explaining each preference decision for auditability
- Track batch status, deadlines, and handoffs between the annotation and QA stages in Jwuma
- Flag ambiguous prompts or rubric gaps to the RLHF program lead
- Maintain consistency scores above program-defined thresholds across sessions
What we look for
- Strong reading comprehension in English; native or professional Uzbek and/or Russian
- Comfortable reasoning about nuanced, sometimes subjective quality judgments
- Reliable, detail-oriented, and able to work to a fixed rubric without personal bias creeping in
- Available onsite at the Samarkand hub on a standard shift schedule
- Basic spreadsheet and web-tool literacy
Nice to have
- Prior exposure to LLMs, chatbots, or AI writing tools
- Background in linguistics, journalism, editing, or research
- Experience with content moderation or editorial QA
Compensation and benefits
- Hourly contract pay, tracked and paid via Jwuma timesheets
- Paid rubric and platform training before live work
- On-site meals/transport stipend where locally applicable
- Route to permanent RLHF Program Associate roles as the Samarkand seat scales
Common questions
What does an RLHF data operations associate do?
They compare AI model outputs against a rubric, record structured preference judgments with a short rationale, and manage batch handoffs — producing the preference data used to align large language models.
Do I need a technical or AI background for this job?
No. Corpshore AI trains new hires on the rubric and the Jwuma platform; strong reading comprehension and sound judgment matter more than a technical background.
Similar roles
Apply once, get matched to what fits
One application covers every role you qualify for. A person reads it.
Apply now