Carlmarx Okemwa

Key Skills

Software

Scale AI

Top Subject Matter

No subject matter listed

Top Data Types

Text

Top Label Types

RLHF

Freelancer Overview

I am a detail-oriented professional with hands-on experience in AI training data collection, video annotation, and remote data entry. My background includes contributing to machine learning datasets by recording and annotating structured task-based videos, ensuring strict adherence to project guidelines and high-quality standards. I excel at following complex instructions, performing repetitive tasks with accuracy, and maintaining consistency across large data cycles. Proficient with tools such as Google Docs, Microsoft Word, Excel, and cloud storage platforms, I am comfortable working independently in remote environments and ensuring precise documentation and data handling. My technical skills and commitment to quality make me well-suited for data labeling and AI data preparation projects.

Labeling Experience

RLHF Expert for STEM & Mathematical Reasoning

Scale AITextRLHF

Project: Large Language Model (LLM) Fine-Tuning & Evaluation Description: I contributed to a major Reinforcement Learning from Human Feedback (RLHF) project aimed at improving the factual accuracy and safety of a conversational AI. My role involved evaluating model-generated responses based on complex rubrics (truthfulness, harmlessness, and helpfulness). I specialized in "Chain-of-Thought" prompting, where I wrote out step-by-step reasoning for mathematical and logical queries to help the model learn structured problem-solving.

Project: Large Language Model (LLM) Fine-Tuning & Evaluation Description: I contributed to a major Reinforcement Learning from Human Feedback (RLHF) project aimed at improving the factual accuracy and safety of a conversational AI. My role involved evaluating model-generated responses based on complex rubrics (truthfulness, harmlessness, and helpfulness). I specialized in "Chain-of-Thought" prompting, where I wrote out step-by-step reasoning for mathematical and logical queries to help the model learn structured problem-solving.

2024 - 2025

Education

Work History