AI Quality Analyst (Personalization): Turkish

What you'll do

Turing's Turkish track for the Gemini personalization project has a distinctive structure: instead of the open-ended "up to 40 hours" of its sibling roles, the ad offers two time-commitment options, 30 hours a week or 40 hours a week. Within that, the work is the same: write multi-turn prompts (typically one to five turns) from your own experiences and judge Turkish replies on grounding, integration and helpfulness.

You rank pairs of responses side by side, justify the ranking in writing with turn references, extract "Debug Info" to confirm which data sources were used, and delete the evaluation chats afterwards.

Who fits

• High-level Turkish reading and writing; Turkish is the project language
• Willing to use your primary personal Google account with personal data sources enabled
• Able to catch subtle naturalness problems and unsupported inferences
• Clear, structured rationales
• Relevant degree or equivalent experience; annotation or AI quality experience strongly preferred
• Desktop or laptop with a good connection

What it pays

$15/hour per the ad, over three months. Dutch and Polish roles pay $20 but run one month.

Worth knowing

Good:

• Predictable volume: 30 or 40 hours a week, your choice
• Three months of work

Less good:

• The minimum is 30 hours a week, less flexible than the siblings that allow as little as 4 hours a day (the ad also says at least 4 hours per day)
• Your personal Google data is the test material
• Full-time availability in your local time zone, with 4 hours of PST overlap
• Contractor, no benefits

How hiring works

Interest form, a 24-hour assessment, then a pre-onboarding discussion.

About this listing

Terms are Turing's; commentary is ours. See Turing.