Sonia Kurniawati

I evaluate alignment between user prompts and AI responses, focusing on noise interference, duration, and strict guideline compliance. Perform verbatim audio transcription under project-specific linguistic standards. Produce high-quality audio data in noisy settings to enhance model robustness and comprehension. Certified Linguistics Specialist in AI-driven tasks.

• Conducted side-by-side evaluations of AI voice agent responses using mobile-based tasks.
• Compared outputs to assess naturalness, clarity, relevance, and linguistic accuracy in Indonesian.
• Contributed to Reinforcement Learning from Human Feedback (RLHF) by providing high-quality
human judgements.
• Identified errors in pronunciation, tone, intent understanding, and conversational flow.

I evaluated AI-generated responses for accuracy, relevance, fluency, and cultural appropriateness in Indonesian and English. Labeled and annotated linguistic data under strict guidelines, provided detailed qualitative feedback to improve model performance, identified errors, ambiguity, and bias in outputs, and maintained high consistency and precision across large datasets.

Grade: 3.6

Activities and societies: • Presenter at the International Conference at UIN Sunan Kalijaga, Yogyakarta, Indonesia • Presenter at the IAFOR International Conference in Tokyo, Japan

Grade: 3.78

Activities and societies: • Student exchange to Asia University, Taiwan • Awardee of Kampus Mengajar Kemendikbudristek Batch 1 • Australian delegate at the International Model United Nations, Malaysia