Karensia Starts

Key Skills

Software

Scale AI

Top Subject Matter

No subject matter listed

Top Data Types

Computer Code Programming

Top Label Types

Computer Programming Coding

Freelancer Overview

I am an experienced AI Trainer and Data Specialist with over 4 years working on large-scale data annotation, LLM evaluation, and AI response quality review projects. My background in computer science and hands-on work with leading platforms like Scale AI and Appen have equipped me to analyze and label thousands of data tasks weekly, focusing on text classification, reasoning validation, and conversational AI. I am skilled at assessing model outputs for accuracy, safety, bias, and contextual alignment, using tools such as Python (Pandas, NumPy), SQL, and advanced Excel. I maintain a strong attention to detail and consistently deliver high-quality results in fully remote, deadline-driven environments, providing structured feedback that improves AI model performance and training pipelines.

Labeling Experience

AI Rater & Content QualityAnalyst

Scale AIComputer Code ProgrammingComputer Programming Coding

Evaluated and labeled 1,000+ data tasks weekly across text classification, reasoning validation, and conversational AI training projects Reviewed and rated AI-generated responses for factual accuracy, coherence, safety compliance, and instruction alignment Identified hallucinations, logical inconsistencies, and bias risks in model outputs Applied detailed annotation guidelines across multi-step reasoning and edge case scenarios Maintained 98% average quality rating across quarterly review cycles Provided structured feedback used to improve model training pipelines

Evaluated and labeled 1,000+ data tasks weekly across text classification, reasoning validation, and conversational AI training projects Reviewed and rated AI-generated responses for factual accuracy, coherence, safety compliance, and instruction alignment Identified hallucinations, logical inconsistencies, and bias risks in model outputs Applied detailed annotation guidelines across multi-step reasoning and edge case scenarios Maintained 98% average quality rating across quarterly review cycles Provided structured feedback used to improve model training pipelines

2021 - 2025

Education

Work History