Derya Yavuz

Evaluated AI-generated Turkish responses against detailed English project rubrics, scoring accuracy, instruction-following, tone, and naturalness, and wrote structured justifications for every rating. Rewrote weak model outputs into idiomatic Turkish and flagged recurring failure patterns to project leads.

Performed RLHF preference ranking on paired Turkish model responses, applying multi-dimensional criteria and documenting deciding factors. Authored gold-standard Turkish prompts and reference responses for supervised fine-tuning and conducted adversarial testing to surface Turkish-specific failure modes.

Reviewed machine-translated and AI-generated Turkish content for accuracy, fluency, and cultural fit, classifying issues by type and severity. Post-edited translated content to publication standard, balancing fidelity to source meaning against natural Turkish phrasing.

Delivered first contract work in AI data and localization, rating AI-generated Turkish responses against client criteria and reviewing machine-translated content for accuracy and fluency. Developed a consistent personal error taxonomy for mistranslation, register, fluency, and terminology.

Bachelor of Science in Computer Engineering from Ondokuz Mayıs University.

Doctor of Veterinary Medicine from Bursa Uludağ University.

Grade: 7.5

IELTS Academic certification with an overall band score of 7.5.