Seohyeon Min

Key Skills

Software

Other

Labelbox

Internal/Proprietary Tooling

Top Subject Matter

No subject matter listed

Top Data Types

Audio

Document

Video

Top Task Types

Classification

Evaluation Rating

Text Generation

Freelancer Overview

I have 6 months of experience in AI training and data labeling at DataFactory, a team of 12 members. During this time, I worked full-time reviewing TTS outputs and AI interview videos, ensuring that responses, facial expressions, eye contact, attire, and applicant profiles met evaluation standards. I also generated interview questions based on applicants’ resumes and verified the quality and appropriateness of these questions, contributing to accurate and effective AI training data. In addition, I participated in developing a Korean spell-check system, collecting large text samples, testing the system’s accuracy, and identifying errors for correction. I conducted monthly quality tests to ensure the system’s reliability. These experiences have strengthened my attention to detail, understanding of language processing, and ability to maintain high-quality AI training data.

Labeling Experience

Native Korean Audio Transcriber -Hangul

LabelboxAudioClassification

-Transcribed Korean audio clips verbatim in Hangul using project style guidelines -Applied annotation rules, speaker formatting, save states, and metadata requirements -Delivered consistent, high-quality batches across multiple days

-Transcribed Korean audio clips verbatim in Hangul using project style guidelines -Applied annotation rules, speaker formatting, save states, and metadata requirements -Delivered consistent, high-quality batches across multiple days

2025 - 2025

Korean Text Annotation & Evaluation (AI Recruitment Data)

Internal Proprietary ToolingTextClassificationText Generation

Reviewed Korean resumes and interview responses to assess linguistic clarity, logical coherence, and question–answer alignment. Labeled response quality based on predefined criteria and added structured annotations for model training. Created supplemental interview questions derived from resume content to support AI-driven question generation. Ensured high accuracy and consistency through strict guideline adherence.

Reviewed Korean resumes and interview responses to assess linguistic clarity, logical coherence, and question–answer alignment. Labeled response quality based on predefined criteria and added structured annotations for model training. Created supplemental interview questions derived from resume content to support AI-driven question generation. Ensured high accuracy and consistency through strict guideline adherence.

2023 - 2024

Korean Video Evaluation for AI Interview Models

Internal Proprietary ToolingVideoClassificationEvaluation Rating

Reviewed Korean interview videos to verify AI-generated assessments related to applicant behavior, gaze direction, potential cheating cues, and overall presentation. Ensured evaluation accuracy by validating model judgments against predefined behavioral criteria and documenting corrections. Labeled video-based attributes to improve multimodal AI interview scoring.

Reviewed Korean interview videos to verify AI-generated assessments related to applicant behavior, gaze direction, potential cheating cues, and overall presentation. Ensured evaluation accuracy by validating model judgments against predefined behavioral criteria and documenting corrections. Labeled video-based attributes to improve multimodal AI interview scoring.

2023 - 2024

Korean Audio Evaluation for STT Model Quality

Internal Proprietary ToolingAudioEvaluation Rating

Evaluated AI-generated Korean STT transcriptions by checking for misrecognitions, spacing/orthography inconsistencies, and guideline deviations. Corrected recognition errors and provided concise reviewer comments when needed. Ensured transcription accuracy and consistency to support high-quality speech recognition dataset development.

Evaluated AI-generated Korean STT transcriptions by checking for misrecognitions, spacing/orthography inconsistencies, and guideline deviations. Corrected recognition errors and provided concise reviewer comments when needed. Ensured transcription accuracy and consistency to support high-quality speech recognition dataset development.

2023 - 2024

Education

Work History