Geraint W.

Key Skills

Software

No software listed

Top Subject Matter

AI model evaluation and safety/quality auditing

Top Data Types

Text

Computer Code Programming

Top Task Types

No task types listed

Freelancer Overview

A.I. Evaluation and Training, Alphabet Inc. Education includes Open-Source Society University - Computer Science Curriculum, Open-Source Society University (2025) and Computation and Programming, MIT Open Courseware (2020). AI-training focus includes data types such as Text and labeling workflows including Evaluation and Rating.

Labeling Experience

A.I. Evaluation and Training, Alphabet Inc. (second stint)

Text

Assessed subjective quality signals for AI responses, scoring conversational metrics such as tone, helpfulness, clarity, and user satisfaction. Verified whether outputs adhered to policy and safety constraints by checking for bias, toxic undertones, policy violations, and subtle hallucinations. Tracked and validated the logical steps an AI agent took to ensure the final answer was grounded in facts rather than guesswork. • Quality scoring of conversational responses • Policy/safety boundary checks (bias, toxicity, violations) • Hallucination detection and edge-case spotting • Reasoning-step validation of agent outputs

2026 - Present

A.I. Evaluation and Training, Alphabet Inc.

Text

Assessed subjective quality signals for AI responses, scoring conversational metrics such as tone, helpfulness, clarity, and user satisfaction. Verified whether outputs adhered to policy and safety constraints by checking for bias, toxic undertones, policy violations, and subtle hallucinations. Tracked and validated the logical steps an AI agent took to ensure the final answer was grounded in facts rather than guesswork. • Quality scoring of conversational responses • Policy/safety boundary checks (bias, toxicity, violations) • Hallucination detection and edge-case spotting • Reasoning-step validation of agent outputs

2025 - 2025

Education

Work History