Hezbon O.

Key Skills

Software

Don't disclose

Other

Top Subject Matter

AI training datasets (NLP, conversational AI, and evaluation/ranking)

LLM prompt-response evaluation for training datasets

Top Data Types

Text

Image

Video

Top Task Types

RLHF

Prompt + Response Writing (SFT)

Bounding Box

Polygon

Segmentation

Evaluation/Rating

Object Detection

Freelancer Overview

AI Trainer & Data Annotator (Freelance/Remote). Brings 6+ years of professional experience across complex professional workflows, research, and quality-focused execution. Core strengths include Don't disclose and Other. Education includes Bachelor of Science, Kibabii University (2021). AI-training focus includes data types such as Text, Computer Code, and Programming and labeling workflows including RLHF and Prompt + Response Writing (SFT).

Labeling Experience

AI Trainer & Data Annotator (Freelance/Remote)

Don't discloseTextRLHF

Provided high-quality AI training data annotations for NLP, computer vision, and conversational AI datasets across remote platforms. Evaluated AI-generated responses for accuracy, coherence, and safety using RLHF (Reinforcement Learning from Human Feedback) style assessment. Worked to improve model response quality by reviewing, ranking, and stress-testing prompts for edge cases. • Conducted response evaluation for helpfulness, coherence, and safety • Reviewed and ranked model outputs to refine tone and factual correctness • Participated in guideline-based annotation to maintain label consistency • Collaborated with remote teams to achieve high inter-annotator agreement.

2022 - Present

AI Prompt & Response Evaluation Tool (Personal Project)

OtherPrompt + Response Writing (SFT)

Built an internal prompt-response evaluation web tool to support AI training dataset creation and quality scoring. Implemented a scoring rubric to rate prompt-response pairs using the HHH (helpfulness, harmlessness, honesty) framework. Used the application to categorize and rate examples for subsequent model training and evaluation workflows. • Collected and categorized prompt-response pairs for training use • Applied HHH scoring (helpfulness, harmlessness, honesty) • Generated ratings to assess response quality for dataset creation • Structured an evaluation workflow for prompt-response datasets.

2023 - 2023

Education

Work History