Generalist AI Trainer at AI Startup

Lightly AG is a Zurich-based AI company and ETH/HSG spin-off, backed by Y Combinator and top-tier investors. Our machine learning and computer vision technology is trusted by global leaders in autonomous driving, medical imaging, and visual inspection.

As we push the boundaries of what AI can do, we need sharp, curious people to help us make it better. We're looking for Generalist AI Trainers to join us on a part-time contractor basis — no technical background required, just strong judgment, clear thinking, and an eye for quality.

The minimum expected time commitment is 10 hours/week.

Tasks

What you'll be doing

• Review and rate AI-generated responses for quality, accuracy, and helpfulness
• Write prompts, questions, and example conversations to help train the model
• Compare pairs of AI responses and select the better one based on clear criteria
• Follow detailed instructions to complete structured evaluation tasks
• Flag problematic or incorrect outputs and explain what's wrong and why
• Occasionally research topics to verify whether AI responses are factually sound

Requirements

What we're looking for

• Strong written English communication skills — you can express your reasoning clearly and concisely
• Sharp attention to detail and consistent judgment across repetitive tasks
• Ability to work independently, manage your own time, and meet deadlines
• Comfortable with ambiguity — not every task has a clear-cut answer
• Reliable internet connection and a quiet working environment
• Availability for at least 10 hours/week on a flexible schedule

Nice to have

• Background in any domain is welcome — we are actively upgrading our skillsets and our offerings
• Experience with AI tools, data annotation, or content evaluation
• Fluency in additional languages beyond English

Benefits

• Competitive compensation tailored to experience and scope of work

Send us your CV along with a brief note about yourself. If you've done similar work before — evaluating AI outputs, annotation, content review, or anything adjacent — we'd love to hear specific examples.

We look forward to hearing from you!