For employers

Hire this AI Trainer

Sign in or create an account to invite AI Trainers to your job.

Invite to Job
Takumi Shiota

Takumi Shiota

LLM Evaluation, Text Generation, Task Reviewing in Japanese & English

Japan flagYamagata, Japan
$30.00/hrExpertData Annotation TechRemotasksOther

Key Skills

Software

Data Annotation TechData Annotation Tech
RemotasksRemotasks
Other
Scale AIScale AI

Top Subject Matter

No subject matter listed

Top Data Types

AudioAudio
DocumentDocument
TextText

Top Task Types

Evaluation Rating
Prompt Response Writing SFT
RLHF

Freelancer Overview

Experiencing variety of LLM assessing, generating, and reviewing tasks like fact-checking, prompt-creation and assessment of responses, editing responses, assessing harmful prompt & responses, creating fine-grained criteria for assessing the quality of responses, reviewing other contributors' tasks, and else.

ExpertEnglishJapanese

Labeling Experience

Data Annotation Tech

Response comparison

Data Annotation TechTextRLHF
Assessing the quality of two to three response simultaneously, rate them, and decide the preference.

Assessing the quality of two to three response simultaneously, rate them, and decide the preference.

2024
Scale AI

Review of other contributors' tasks

Scale AITextRLHF
Reviewing the works completed by the other contributors, before actually confirm and send them to the client. Highly important as it is for ensuring the quality of each tasks.

Reviewing the works completed by the other contributors, before actually confirm and send them to the client. Highly important as it is for ensuring the quality of each tasks.

2022
Scale AI

Transcription

Scale AIAudioRLHFData Collection
Listen to an audio whose length is a minute at maximum, and create the transcript for it. I've been both a tasker and a reviewer

Listen to an audio whose length is a minute at maximum, and create the transcript for it. I've been both a tasker and a reviewer

2025 - 2025
Data Annotation Tech

Harmful prompt and response

Data Annotation TechTextRLHF
Handling prompts and responses that could potentially hurt or distress the users. It measures and categorise them according to the safety categories.

Handling prompts and responses that could potentially hurt or distress the users. It measures and categorise them according to the safety categories.

2024 - 2025
Data Annotation Tech

Fine-grained criteria

Data Annotation TechTextRLHF
Crafting a criteria which is used to assess the quality of the AI responses. Like, it assesses if the responses follow the instruction, generate factually correct answer, sound naturally, care about the design of the answer, and else.

Crafting a criteria which is used to assess the quality of the AI responses. Like, it assesses if the responses follow the instruction, generate factually correct answer, sound naturally, care about the design of the answer, and else.

2024 - 2025

Education

T

Toyo University

Bachelor's degree in Global Innovation Studies, Global Innovation Studies

Bachelor's degree in Global Innovation Studies
2017 - 2021

Work History

F

FLAT Bouldering

Gym Staff

Yamagata
2023 - Present
N

Natu Bouldering

Content Creator

Tokyo
2022 - Present