Work on the data behind frontier AI.

Rate, rank and red-team model outputs across reasoning, coding, safety and instruction-following rubrics. Ideal for engineers, researchers and strong generalists.

Audit translations, generation quality and dialectal nuance in Swahili, Yoruba, Amharic, Hausa, Zulu, Wolof and 30+ languages.

Produce bounding boxes, segmentation masks and keypoints on street scenes, agriculture, healthcare and retail imagery with 99%+ IoU targets.

Apply your professional expertise in medicine, law, finance or engineering to evaluate model outputs, write guidelines and catch safety-critical errors.

Probe models for harmful, biased or misleading outputs and document reproducible harm taxonomies. Strong writing and structured judgment required.

Own delivery for customer projects: roster management, quality telemetry, reviewer coaching and daily client communication.

Evaluate model responses against safety policies, flag edge-case harms and write concise justifications that improve guardrails. Similar to specialist evaluator roles on top AI labs' partner platforms.

Review, rank and explain code generations across Python, JavaScript, C++ and Go. Ideal for early-career SWEs, bootcamp grads and CS students who can spot subtle bugs.

Verify factual claims, source quality and tone in generated content. Flag hallucinations, outdated references and unsupported assertions for search, chat and research products.

Compare model-written drafts, rank clarity and style, and rewrite weak outputs to train more helpful, human-like writers. Strong C1+ English and editing experience preferred.

Evaluate ML, statistics and systems-design responses; write rubrics and gold explanations that teach models to reason like senior data scientists.

A broad entry point for sharp generalists: compare answers, follow complex instructions, spot inconsistencies and provide readable explanations across many topics.