Sungho Kim

Key Skills

Software

Other

Top Subject Matter

No subject matter listed

Top Data Types

Computer Code Programming

Document

Text

Top Task Types

Classification

Computer Programming Coding

RLHF

Text Generation

Text Summarization

Freelancer Overview

With a strong specialization in evaluating Korean and English LLMs, I have gained extensive experience in Reinforcement Learning from Human Feedback (RLHF) through my work on the Outlier.ai platform. My expertise includes designing and assessing responses to improve AI performance, with a focus on critical criteria such as conciseness, factual accuracy, adherence to instructions, and maintaining an appropriate tone. One of my key projects involved creating intentionally flawed responses related to Korean traditional culture, historical facts, transportation, and geographical features to identify specific areas where AI models failed. By analyzing these failures, I provided valuable insights to enhance model performance. Additionally, I conducted detailed evaluations of RLHF assessments performed by others, assessing the quality and completeness of their reviews. These tasks required a meticulous approach to identify shortcomings in the evaluations and ensure alignment with RLHF goals. This diverse experience has equipped me with a deep understanding of AI training processes and the ability to contribute meaningfully to the development and refinement of LLMs.

Labeling Experience

Education

Work History