Open to opportunities

Ferhat Tunc

@tncdeveloper

Message

I evaluate Turkish AI responses for accuracy, safety, and language quality.

Turkey

Message

What I'm looking for

I'm looking for remote AI evaluation, localization, content quality, or trust-and-safety work where I can apply native Turkish and Kurdish language judgment, structured rubrics, fact-checking, and AI model testing.

I've evaluated AI chatbot responses, reasoning, and training data for Outlier, RWS/Multimango, Appen/CrowdGen, OneForma, and Centific across thousands of conversation turns.

I detect hallucinations, harmful content, and subtle quality failures in Turkish-language AI outputs, including responses that appear correct but are not. My rubric-based evaluations reduced harmful or inefficient outputs by 40%+ and improved dataset accuracy across 500+ model-training tasks.

I work across Turkish, Kurdish, English, Azerbaijani, and Finnish content, handling translation QA, OCR and transcription annotation, fact-checking, and trust-and-safety classification. I bring native Turkish and Kurdish language insight, including regional slang, cultural context, tone, and fluency.

Alongside my freelance AI evaluation work, I'm studying Computer Engineering at Kocaeli University and use SQL, Python, Java, and AI tools including Claude, ChatGPT, and GitHub Copilot in structured evaluation workflows.

Experience

Work history, roles, and key accomplishments

Current

AI Content Evaluation & Localization Specialist

Current

Outlier, RWS/Multimango, Appen/CrowdGen, OneForma, Centific

Jan 2023 - Present (3 years 7 months)

Evaluated and ranked AI chatbot responses against structured rubrics, detecting hallucinations and safety issues in Turkish and Kurdish content. Applied rubric-based error categorization for translation QA and annotated multilingual data for OCR and transcription tasks.