AI Training Analyst

AI Training Analyst for research, evaluation & data quality (remote, contract)
Location: 100% remote (global); must be able to overlap at least 4 hours/day with US Pacific Time
Engagement: Independent contractor / freelance; project-based (typically 2–4 months, extendable based on performance)
Commitment: 20 to 40 hours per week (minimum 4 hours/day)
Compensation: Hourly or per-task, varies by project; typical range $15–$30 per hour or equivalent
About the opportunity
We are recruiting for a fast-growing San Francisco AI company that partners with the world's leading AI labs (including Google Gemini) to improve frontier models. The work is not engineering: it is human judgment at scale. As an AI Training Analyst you will research, reason, evaluate and annotate so that AI systems become more accurate, helpful and safe. Projects are short, fully remote and assessment-based, which makes this a good fit for students, working professionals, writers, researchers and analysts who want hands-on exposure to how modern AI is trained. Every specific project assignment will depend on your background, availability and languages.
Why join
- Work remotely and independently on short-term AI training projects, with a flexible hours commitment that fits around studies or another job
- No engineering or specialized AI background required: what matters is attention to detail, sound judgment and clear writing. You receive detailed guidelines and onboarding, and complete a single assessment before you start
- Get a first-hand look at how frontier AI labs train and evaluate their models, and get paid in USD while you learn
- Strong performance on one project makes you eligible for follow-on projects and extensions, so a short engagement often turns into ongoing work
What you'll do
- Research and verify information from online sources, analyze complex content and logical problems, and summarize findings into clear, concise insights
- Solve analytical and reasoning tasks (data interpretation, logic puzzles, basic quantitative problems) and write the correct answer and explanation so the model can learn from it
- Design prompts and realistic scenarios that test AI models, including multi-turn conversations, and evaluate the quality of the responses they produce
- Review AI-generated responses for accuracy, grounding, relevance and helpfulness; compare responses side by side and rank them
- Write clear, structured, defensible rationales for every judgment, referencing the specific evidence or conversation turn that supports it
- Follow detailed annotation guidelines precisely, flag edge cases, and maintain consistency and quality across a high volume of tasks
What we're looking for
- Strong English comprehension and writing skills (professional level or higher)
- Sharp analytical and critical-thinking skills, with the ability to evaluate nuanced or ambiguous content and spot subtle errors
- Meticulous attention to detail and the discipline to follow structured guidelines
- Excellent written communication: explaining why something is right or wrong is the core of the job
- Self-motivated and able to work independently in a remote setting, meeting quality and throughput targets
- Availability of at least 20 hours per week with a 4-hour daily overlap with US Pacific Time, and a reliable computer with a stable internet connection
Nice to have
- Fluent Spanish (unlocks projects that are bilingual English/Spanish or Spanish-focused)
- A bachelor's degree, completed or in progress, or equivalent experience in research, analysis, writing, editing, translation, journalism, linguistics, policy, law or data
- Prior experience in data annotation, AI quality evaluation, content moderation or LLM evaluation
- Professional writing experience (analyst, copywriter, journalist, technical writer, editor, translator)
- Investigative research skills: locating primary sources, navigating archives, databases and registries, fact-checking, OSINT or due diligence
- An existing, actively used personal Gemini or Google account (some projects evaluate personalization features and require genuine account history)
- Experience in video/film editing or media production, or familiarity with image and video annotation
- Comfort with Excel/Google Sheets, basic data interpretation and structured formats such as JSON

Work arrangement
Yes