About The Opportunity
Help evaluate how well a new Gemini personalization feature uses a person's past Gemini conversations, Gmail, Google Search, and YouTube activity to make responses more relevant and helpful. You will design realistic multi-turn prompts based on your own experiences, compare model responses, and provide clear evidence-based judgments about grounding, integration, and helpfulness.
What You Will Do
- Design and run creative multi-turn prompts, typically one to five turns, that require the model to use personal context.
- Compare two model responses side by side and rank their overall quality.
- Identify unsupported claims, incorrect personalization, weak inferences, forced connections, and unnatural overexplaining.
- Write concise, defensible rationales that point to the relevant turns and evidence.
- Review debug information to verify that summaries and authorized data sources were used correctly.
- Follow project data-hygiene instructions, including deleting evaluation conversations when required.
What We Are Looking For
- Strong Chinese reading and writing proficiency.
- At least one year of relevant professional or equivalent experience.
- A bachelor's degree or equivalent experience in a relevant analytical field, such as policy, law, ethics, linguistics, journalism, computer science, or a related field.
- Strong analytical judgment, attention to detail, written communication, and the ability to work independently.
- Experience with data annotation, AI quality evaluation, content moderation, prompt design, or similar analytical work is helpful.
- Willingness to use your primary personal Google account, not a testing account, and enable approved personal data sources for a genuine assessment.
- A desktop or laptop, reliable internet connection, and a suitable remote work setup.
Engagement details
- Remote contractor assignment with an immediate start and an expected duration of four months.
- Full-time availability in your local time zone, with 30 or 40 hours per week and at least four hours of overlap with Pacific Time.
- Open to candidates physically located in Malaysia, Thailand, Singapore, or an approved U.S. state.
Compensation
