佐藤 駿平

Key Skills

Software

Appen

Clickworker

Data Annotation Tech

Labelbox

Mercor

OneForma

Scale AI

Top Subject Matter

No subject matter listed

Top Data Types

Audio

Document

Image

Top Task Types

Audio Recording

Evaluation Rating

Prompt Response Writing SFT

Text Generation

Text Summarization

Freelancer Overview

As an expert in data labeling and AI training data creation, I have contributed to mathematics, image-based, and audio transcription projects for various clients. I have designed task-optimized prompts and performed high-accuracy annotations across multiple data types such as the above, ensuring consistency, precision, and adherence to rigorous quality standards. My multilingual background and experience in cross-domain AI annotation projects set me apart, enabling me to understand model behavior deeply and evaluate outputs with both technical accuracy and linguistic nuance. With two years of hands-on experience, I bring reliability, analytical insight, and a strong commitment to producing high-quality training data.

Labeling Experience

Rating between 2 audios

ClickworkerAudioEvaluation Rating

Rating between an audio with a same script with a different speaker.

Rating between an audio with a same script with a different speaker.

2025

Transcription, rating of audio

LabelboxAudioAudio Recording

A project where you listen to a 30-second audio file, transcribe it, and rate the speaker's sound quality and content.

A project where you listen to a 30-second audio file, transcribe it, and rate the speaker's sound quality and content.

2025

General domain prompting and evaluation, labeling of the response

MercorTextText Generation

Created prompts spanning various domains and evaluated the quality of the model's responses.

Created prompts spanning various domains and evaluated the quality of the model's responses.

2025 - 2025

Multi-person conversation transcription & labeling

Scale AIAudioClassificationText Generation

Transcribed and labeled multi-person conversational audio data, including speaker diarization and possibly sentiment labeling.

Transcribed and labeled multi-person conversational audio data, including speaker diarization and possibly sentiment labeling.

2025 - 2025

Harmful content prompt creation & safety rating

Scale AITextEntity Ner ClassificationText Generation

Created prompts that could lead to harmful outputs, rated the risk level of the resulting responses to train response models.

Created prompts that could lead to harmful outputs, rated the risk level of the resulting responses to train response models.

2025 - 2025

Show more

Education

Work History