Samuel Mutui

Key Skills

Software

Prodigy

Top Subject Matter

No subject matter listed

Top Data Types

Text

Top Label Types

Segmentation

Classification

Freelancer Overview

I am an experienced AI training and prompt engineering specialist with a strong background in creating, refining, and evaluating high-quality training data for large language models. My expertise includes crafting clear and accurate prompts, performing critical evaluation and adversarial testing of AI outputs, and conducting thorough fact-checking and technical research to ensure data integrity. I have worked across diverse domains, consistently mastering complex project guidelines to deliver precise, impactful results that improve AI model performance and safety. With advanced English writing skills, a detail-oriented approach, and a proven ability to work independently and meet deadlines, I am dedicated to producing reliable and effective training data for cutting-edge AI systems.

Labeling Experience

Sentiment Analysis on Customer Reviews for E-commerce

ProdigyTextSegmentationClassification

Led a large-scale data labeling initiative focused on classifying sentiment (positive, neutral, negative) and identifying user intent (purchase inquiry, complaint, feedback, support request) from e-commerce customer reviews and chat logs. Managed a dataset of over 500,000 text entries, ensuring labeling consistency and accuracy across a team of 12 annotators. Implemented a multi-stage QA process including inter-annotator agreement checks and regular review cycles to maintain >95% label accuracy. Worked closely with NLP engineers to refine label taxonomy and improve model performance over time.

Led a large-scale data labeling initiative focused on classifying sentiment (positive, neutral, negative) and identifying user intent (purchase inquiry, complaint, feedback, support request) from e-commerce customer reviews and chat logs. Managed a dataset of over 500,000 text entries, ensuring labeling consistency and accuracy across a team of 12 annotators. Implemented a multi-stage QA process including inter-annotator agreement checks and regular review cycles to maintain >95% label accuracy. Worked closely with NLP engineers to refine label taxonomy and improve model performance over time.

2022 - 2024

Education

Work History