Looks like you have JavaScript disabled. For the full Twine experience, you will need to re-enable it.

I’m Catherine Warugu, an AI Data Annotation & Evaluation Specialist and English content writer based in Kenya, currently supporting AI training workflows as a Data Trainer at SuperAnnotate. I help teams build high-quality labeled datasets by applying detailed guidelines, catching systematic labeling issues early, and improving inter-annotator consistency. Alongside evaluation and QA work, I’m also a creative and structured writer—turning complex ideas into clear content and polished writing. I’ve developed annotation decision trees for ambiguous cases, evaluated LLM outputs for safety and hallucinations, and created prompt-response pairs that improve training reliability. My goal is always the same: accurate data, consistent standards, and communication that’s easy to understand.…I’m Catherine Warugu, an AI Data Annotation & Evaluation Specialist and English content writer based in Kenya, currently supporting AI training workflows as a Data Trainer at SuperAnnotate. I help teams build high-quality labeled datasets by applying detailed guidelines, catching systematic labeling issues early, and improving inter-annotator consistency. Alongside evaluation and QA work, I’m also a creative and structured writer—turning complex ideas into clear content and polished writing. I’ve developed annotation decision trees for ambiguous cases, evaluated LLM outputs for safety and hallucinations, and created prompt-response pairs that improve training reliability. My goal is always the same: accurate data, consistent standards, and communication that’s easy to understand.

Catherine Warugu

AI Evaluator, Data Annotator, AI Trainer, +1





I’m Catherine Warugu, an AI Data Annotation & Evaluation Specialist and English content writer based in Kenya, currently supporting AI training workflows as a Data Trainer at SuperAnnotate. I help teams build high-quality labeled datasets by applying detailed guidelines, catching systematic labeling issues early, and improving inter-annotator consistency. Alongside evaluation and QA work, I’m also a creative and structured writer—turning complex ideas into clear content and polished writing. I’ve developed annotation decision trees for ambiguous cases, evaluated LLM outputs for safety and hallucinations, and created prompt-response pairs that improve training reliability. My goal is always the same: accurate data, consistent standards, and communication that’s easy to understand.…I’m Catherine Warugu, an AI Data Annotation & Evaluation Specialist and English content writer based in Kenya, currently supporting AI training workflows as a Data Trainer at SuperAnnotate. I help teams build high-quality labeled datasets by applying detailed guidelines, catching systematic labeling issues early, and improving inter-annotator consistency. Alongside evaluation and QA work, I’m also a creative and structured writer—turning complex ideas into clear content and polished writing. I’ve developed annotation decision trees for ambiguous cases, evaluated LLM outputs for safety and hallucinations, and created prompt-response pairs that improve training reliability. My goal is always the same: accurate data, consistent standards, and communication that’s easy to understand.

Available to hire

I’m Catherine Warugu, an AI Data Annotation & Evaluation Specialist and English content writer based in Kenya, currently supporting AI training workflows as a Data Trainer at SuperAnnotate. I help teams build high-quality labeled datasets by applying detailed guidelines, catching systematic labeling issues early, and improving inter-annotator consistency.

Alongside evaluation and QA work, I’m also a creative and structured writer—turning complex ideas into clear content and polished writing. I’ve developed annotation decision trees for ambiguous cases, evaluated LLM outputs for safety and hallucinations, and created prompt-response pairs that improve training reliability. My goal is always the same: accurate data, consistent standards, and communication that’s easy to understand.

Skills

Experience Level

Expert

Expert

Expert

Expert

Expert

Intermediate

Intermediate

Work Experience

Data Trainer at SuperAnnotate

October 1, 2026 - Present

Contribute to AI data training and annotation workflows, applying task-specific guidelines accurately. Evaluate data quality and support accurate, consistent, high-quality labeled data. Apply analytical reasoning and attention to detail across AI training tasks.

AI Data Annotation & Evaluation Specialist at BELAY

January 1, 2023 - December 31, 2025

Delivered high-fidelity labeled datasets at scale, including 200+ images annotated with 2D bounding boxes in CVAT and maintaining 98%+ accuracy. Reviewed and validated 300+ text and image annotations, correcting labeling errors to protect model performance. Developed annotation decision trees for ambiguous cases, reducing inter-annotator disagreement by 25% and increasing labeling speed by 20%. Evaluated LLM outputs against safety policies by flagging hallucinations and policy violations with a 95%+ detection rate. Implemented systematic error-detection processes that reduced dataset rework by 40% and saved 200+ hours of re-annotation effort. Onboarded quickly by mastering task-specific guidelines within 48 hours and meeting quality benchmarks across 5+ concurrent projects with shifting requirements.

Customer Service & Data Support at Brickman Properties Limited

June 1, 2022 - September 30, 2022

Resolved 100+ customer disputes with a 95%+ satisfaction rate using structured problem-solving and clear communication. Maintained 100% accuracy on communication records and data entry across high-volume interactions. Built systems for tracking and organizing customer data with zero errors over a four-month tenure.

Records and Data Support Intern at Ministry of Lands, Housing and Urban Development, Thika

January 1, 2017 - January 1, 2018

Processed and validated 300+ title deed records with 100% accuracy, ensuring compliance and data integrity. Organized and indexed 500+ documents, reducing data retrieval time by 50%.