Ahra I.

Every AI model is only as smart as the data it was trained on. I make that data exceptional. I'm an AI Data Specialist with 5+ years of focused experience in LLM training, data annotation, and multilingual dataset creation. I work across English and Korean at a professional level, with additional coverage in Tamil, Sinhala, and Hindi. Over the years I've contributed to large-scale AI projects covering a pretty wide range: LLM training data creation (instruction tuning, RLHF, preference ranking, red-teaming, prompt engineering), text and audio annotation, handwriting and voice dataset curation for speech and OCR models, and iterative feedback loops for model evaluation and improvement. I've also done a lot of document validation, structured data formatting, and QA work that feeds directly into AI pipelines. What actually sets me apart is the language side. Most annotators work in one language. I work across English and Korean natively, which means I can handle bilingual datasets, catch the kind of mistranslations that quietly break model outputs, and produce training data that holds up across both linguistic and cultural contexts. That's a combination that's harder to find than it sounds, and it shows up in my client reviews. On top of the annotation work, I spent nearly two years as a Project Manager and QA lead at a Korean IT company. I managed cross-functional teams, ran QA processes, wrote technical documentation, and shipped products. So I understand how annotation work fits into a broader AI development pipeline, not just how to execute the tasks themselves. Tools I've worked with include Doccano, Google Data Compute, and various custom annotation platforms. I pick up new tools quickly and have adapted to client-specific systems without issues. If your project needs high-quality, linguistically nuanced AI training data, especially across English and Korean, I'd love to talk. Plus my initials are AI too, so you could say I was made for this work ;)