Minki Gho

Key Skills

Software

Labelbox

Scale AI

Top Subject Matter

No subject matter listed

Top Data Types

Audio

Image

Text

Top Task Types

Audio Recording

Classification

Evaluation Rating

Question Answering

RLHF

Freelancer Overview

I have hands-on experience in AI training and data labeling, particularly in evaluating and refining AI-generated responses. In my previous role, I was responsible for verifying the factual accuracy of AI outputs and revising them based on reliable and verifiable sources. This work required strong research skills, critical thinking, and attention to detail to ensure that all information aligned with credible evidence and maintained a high standard of accuracy. In addition, I have extensive experience in image labeling and annotation tasks, identifying and categorizing specific elements within images to improve computer vision model performance. Through this work, I developed a solid understanding of data quality, consistency, and annotation guidelines, which are crucial for creating effective AI training datasets. My combined experience in text-based and visual data evaluation has equipped me with a well-rounded perspective on AI model development and optimization.

Labeling Experience

contributor

Scale AITextEvaluation RatingPrompt Response Writing SFT

I was responsible for verifying the factual accuracy of AI-generated responses and revising them when necessary, using reliable sources as evidence. This task was conducted roughly every two months, and quality management involved ensuring compliance with provided guidelines as well as incorporating feedback from reviewers. In addition, I designed prompts and rubrics to intentionally induce model failures, evaluated the resulting flawed responses according to my rubric, and then refined them for improvement. I also worked on labeling images by identifying and tagging specific components within them. This was part of a three-month project, during which I followed the company’s detailed guidelines. For example, I avoided creating overly simple prompts or rubrics, and steered clear of trivial, quiz-like questions. Throughout the project, my work was regularly reviewed by other reviewers who provided constructive feedback.

I was responsible for verifying the factual accuracy of AI-generated responses and revising them when necessary, using reliable sources as evidence. This task was conducted roughly every two months, and quality management involved ensuring compliance with provided guidelines as well as incorporating feedback from reviewers. In addition, I designed prompts and rubrics to intentionally induce model failures, evaluated the resulting flawed responses according to my rubric, and then refined them for improvement. I also worked on labeling images by identifying and tagging specific components within them. This was part of a three-month project, during which I followed the company’s detailed guidelines. For example, I avoided creating overly simple prompts or rubrics, and steered clear of trivial, quiz-like questions. Throughout the project, my work was regularly reviewed by other reviewers who provided constructive feedback.

2023 - 2025

Education

Work History