Judith Ndunya J.

Key Skills

Software

Don't disclose

Top Subject Matter

Cultural authenticity assessment for AI training datasets (health content

Luo community customs)

AI output evaluation and quality assurance for business analysis/data tasks

Top Data Types

Text

Document

Top Task Types

Prompt + Response Writing (SFT)

Freelancer Overview

AI training specialist with hands-on experience in data annotation, cultural authenticity evaluation, and model output quality assurance. At Microsoft, I evaluated AI-generated content for cultural accuracy using a structured 0–100 scoring rubric, flagging inaccuracies and providing detailed corrections to improve model outputs. At Turing, I assessed AI responses for logical reasoning and accuracy, contributing to prompt evaluation and quality assurance workflows. Across both roles, I delivered precise, structured feedback at high volume within defined quality frameworks.

Labeling Experience

Business Analyst / AI Contractor — Turing (Contract)

Don't discloseTextPrompt + Response Writing (SFT)

Collaborated with AI product teams to evaluate, test, and improve quality of AI-generated outputs across business analysis and data tasks. Analyzed model responses for accuracy, logical reasoning, and alignment with business requirements, and provided structured feedback for model refinement. Supported prompt evaluation and quality assurance to ensure outputs met defined enterprise performance standards.•Reviewed and assessed AI responses for correctness and reasoning quality.•Compared outputs to business requirements and provided structured improvement feedback.•Contributed to prompt evaluation and QA processes for enterprise use cases.•Benchmarked AI capabilities against real-world analyst workflows.

2026 - 2026

Data Annotator (AI Training) — Microsoft (Contract)

Don't discloseTextEvaluation/Rating

Evaluated AI-generated health narratives for cultural authenticity and accuracy using domain knowledge of Luo customs and language patterns. Applied a structured 0–100 rubric at the paragraph and story level to identify authentic cultural markers and flag inaccuracies. Produced detailed corrective feedback to improve future AI model outputs for East African cultural representation.•Assessed paragraph-level and story-level cultural authenticity using a 0–100 scoring rubric.•Identified authentic markers such as food, music, geography, and occupations.•Flagged non-native terms and factual or cultural inconsistencies.•Provided written feedback and corrections to guide model refinement.

2026 - 2026

Education

Work History