I will annotate and label japanese text and audio data for ai training
Native Japanese AI Evaluator, LLM Data Annotation and AI Automation
Over deze dienst
tom offer. Based in Tokyo (JST), English business level.Native Japanese data annotation for AI and ML teams, done by someone who has worked as an evaluator for Invisible Technologies, Scale AI, RWS, and Turing since 2024.
What I can label
- Text: classification, sentiment, intent, entity tagging (NER), spans, relevance, toxicity and safety flags
- LLM outputs: rubric scoring, pairwise ranking, factuality and instruction-following checks, written rationales
- Audio: transcription review, word-boundary checks, speaker and quality tags
- Images: captions, bounding-box QA, and review of Japanese text in images
How I work
- I follow your guideline exactly and log ambiguous cases instead of guessing
- Consistent labels across the whole batch, with a self-QA pass before delivery
- Feedback on unclear guideline rules so your next batch is cleaner
- Familiar with Label Studio, Argos-style web tools, Google Sheets, CSV, and JSONL
What you get
Techniek:
Handleiding
Type tagging:
Tekst
•
Afbeelding
•
Audio
Veelgestelde vragen
What counts as one item?
One text sample, one LLM response (or one pair for ranking), one audio clip up to 30 seconds, or one image. If your items are much longer or need several labels each, send a sample and I will quote.
Can you work inside our own annotation tool?
Yes. I regularly work in vendor tools such as Label Studio and Argos-style web platforms. Send an invite and the guideline, and I will complete any qualification test first.
