Arisa Otsuka

Key Skills

Software

OneForma

Scale AI

Top Subject Matter

No subject matter listed

Top Data Types

Computer Code Programming

Geospatial Tiled Imagery

Image

Top Task Types

Classification

Computer Programming Coding

Data Collection

Fine Tuning

Segmentation

Freelancer Overview

I am a Data Science & AI Strategy MSc student with 2.5+ years’ experience in analytics, AI training, and human-in-the-loop model evaluation. My expertise spans LLM prompt quality assessment, multilingual data labeling (English, Japanese, French, Mandarin), and model evaluation across NLP, data science, and structured datasets. I have reviewed and QA-tested over 150 AI prompts for multilingual models, achieving a 98% acceptance rate and overall 4.7/5 quality score. My work includes designing annotation guidelines, coding in Python and LaTeX, curating datasets, and evaluating model outputs for accuracy, coherence, and compliance with requirements. Alongside AI training, I bring strong technical skills in Python, SQL, statistical analysis, and data visualization (Power BI, Tableau, Looker), enabling me to support data preparation, feature engineering, and exploratory data analysis for AI workflows. I have contributed to projects across EdTech, FinTech, and media analytics, working effectively with cross-functional teams to deliver high-quality, production-ready data assets.

Labeling Experience

Multilingual LLM Prompt & Code Evaluation (Python, LaTeX)

Scale AIComputer Code ProgrammingQuestion AnsweringText Generation

Reviewed and QA-tested 150+ AI prompts for multilingual models (English, Japanese, French, Mandarin) involving Python-based math logic and LaTeX formatting. Designed annotation guidelines, curated datasets, and evaluated model outputs for accuracy, coherence, and compliance with project requirements, achieving a 98% acceptance rate and 4.7/5 quality score.

Reviewed and QA-tested 150+ AI prompts for multilingual models (English, Japanese, French, Mandarin) involving Python-based math logic and LaTeX formatting. Designed annotation guidelines, curated datasets, and evaluated model outputs for accuracy, coherence, and compliance with project requirements, achieving a 98% acceptance rate and 4.7/5 quality score.

2024 - 2025

Education

Work History