Looks like you have JavaScript disabled. For the full Twine experience, you will need to re-enable it.

AI data annotation, evaluation, and localization specialist with 2+ years of experience across 5+ freelance platforms, delivering dataset labeling for text, audio, image/OCR, and LLM safety tasks. Native Turkish & Kurdish, with C2-level professional English, I apply rubric-driven QA to reduce mistranslations, hallucinations, and harmful outputs. I work fully remote and asynchronously, using Python/SQL and structured validation workflows to ensure high inter-rater agreement and accuracy across multilingual projects spanning regional slang and cultural nuance.…AI data annotation, evaluation, and localization specialist with 2+ years of experience across 5+ freelance platforms, delivering dataset labeling for text, audio, image/OCR, and LLM safety tasks. Native Turkish & Kurdish, with C2-level professional English, I apply rubric-driven QA to reduce mistranslations, hallucinations, and harmful outputs. I work fully remote and asynchronously, using Python/SQL and structured validation workflows to ensure high inter-rater agreement and accuracy across multilingual projects spanning regional slang and cultural nuance.

Ferhat Tunç

Data Annotator, Data Scientist, Programmer, +1





AI data annotation, evaluation, and localization specialist with 2+ years of experience across 5+ freelance platforms, delivering dataset labeling for text, audio, image/OCR, and LLM safety tasks. Native Turkish & Kurdish, with C2-level professional English, I apply rubric-driven QA to reduce mistranslations, hallucinations, and harmful outputs. I work fully remote and asynchronously, using Python/SQL and structured validation workflows to ensure high inter-rater agreement and accuracy across multilingual projects spanning regional slang and cultural nuance.…AI data annotation, evaluation, and localization specialist with 2+ years of experience across 5+ freelance platforms, delivering dataset labeling for text, audio, image/OCR, and LLM safety tasks. Native Turkish & Kurdish, with C2-level professional English, I apply rubric-driven QA to reduce mistranslations, hallucinations, and harmful outputs. I work fully remote and asynchronously, using Python/SQL and structured validation workflows to ensure high inter-rater agreement and accuracy across multilingual projects spanning regional slang and cultural nuance.

Available to hire

AI data annotation, evaluation, and localization specialist with 2+ years of experience across 5+ freelance platforms, delivering dataset labeling for text, audio, image/OCR, and LLM safety tasks. Native Turkish & Kurdish, with C2-level professional English, I apply rubric-driven QA to reduce mistranslations, hallucinations, and harmful outputs.

I work fully remote and asynchronously, using Python/SQL and structured validation workflows to ensure high inter-rater agreement and accuracy across multilingual projects spanning regional slang and cultural nuance.

Skills

Experience Level

Expert

Expert

Expert

Intermediate

Intermediate

Intermediate

Intermediate

Intermediate

Intermediate

Beginner

Beginner

Beginner

Beginner

Language

Turkish

Fluent

Kurdish

Fluent

English

Fluent

Azerbaijani

Advanced

Uzbek

Advanced

Polish

Intermediate

German

Intermediate

French

Intermediate

Catalan; Valencian

Intermediate

Spanish; Castilian

Intermediate

Work Experience

AI Data Annotation, Evaluation & Localization Specialist at Freelance (Outlier, RWS/Multimango, Appen/CrowdGen, OneForma, Centific)

January 1, 2023 - Present

Freelance specialist for multilingual AI data annotation and evaluation across multiple platforms. Perform translation quality assessment (error categorization such as omission/addition/mistranslation/fluency), OCR/document annotation, audio/TTS evaluation and transcription, and content verification with fact-checking for verification accuracy. Contribute to LLM training data work by ranking model outputs against RLHF safety rubrics and applying trust & safety risk-mitigation classification to identify harmful content types. Improve dataset accuracy by detecting hallucinations and verifying claims using SQL-based checks. Standardize labeling consistency across distributed teams via multi-tier acceptability rubrics to improve inter-rater agreement. Maintain high task completion reliability using Python/SQL scripting for structured dataset QA.