Andrew King

Key Skills

Software

SuperAnnotate

Top Subject Matter

No subject matter listed

Top Data Types

Text

Top Label Types

Evaluation Rating

Red Teaming

RLHF

Freelancer Overview

With extensive experience architecting secure AI and cloud platforms for leading organisations, I bring a deep understanding of data governance, AI security posture management (AI-SPM), and the practical application of responsible AI principles. My background includes hands-on work with data labeling, annotation, and AI training data pipelines, particularly within Microsoft Azure AI, OpenAI, and Kubernetes environments. I have led projects involving AI threat modelling, data security compliance, and red-teaming for LLMs, ensuring high-quality, compliant, and robust training datasets. My expertise spans tools such as PyRIT, Guardrails-AI, DeepEval, and Microsoft Purview, supporting both structured and unstructured data workflows for NLP and generative AI use cases. I am passionate about enabling secure, scalable, and ethical AI solutions by building and optimizing data pipelines, collaborating closely with engineering and governance teams, and applying industry best practices to every stage of the AI data lifecycle.

Labeling Experience

AI Evaluation Trainer - Safety Red Teamer Specialist

SuperannotateTextRed Teaming

NDA

NDA

2025 - 2025

AI Evaluation Trainer

TextEvaluation Rating

As an AI Evaluation Trainer for Frontier Model, I was tasked with ensuring the safety and reliability of AI-generated content. My role involved using harm taxonomies to identify biases, misinformation, unsafe reasoning, and disallowed content. I consistently documented clear reasoning to enhance internal guidelines and evaluation frameworks. • Evaluated AI-generated text for biases and misinformation. • Contributed high-quality human data for AI safety research. • Collaborated asynchronously with research teams to improve safety standards. • Identified subtle unsafe behaviors missed by automated systems.

As an AI Evaluation Trainer for Frontier Model, I was tasked with ensuring the safety and reliability of AI-generated content. My role involved using harm taxonomies to identify biases, misinformation, unsafe reasoning, and disallowed content. I consistently documented clear reasoning to enhance internal guidelines and evaluation frameworks. • Evaluated AI-generated text for biases and misinformation. • Contributed high-quality human data for AI safety research. • Collaborated asynchronously with research teams to improve safety standards. • Identified subtle unsafe behaviors missed by automated systems.

Not specified

AI Evaluation Trainer - Safety Red Teamer Specialist

TextEvaluation Rating

In this role, I was responsible for task tagging, annotating, and evaluating AI-generated text against safety criteria. This included assessing aspects like bias, misinformation, unsafe reasoning, and disallowed content. I applied harm taxonomies consistently, even in ambiguous scenarios. • Identified and documented subtle unsafe behaviors. • Contributed high-quality human data for AI safety research. • Collaborated asynchronously with research and safety teams. • Documented reasoning to enhance evaluation frameworks.

In this role, I was responsible for task tagging, annotating, and evaluating AI-generated text against safety criteria. This included assessing aspects like bias, misinformation, unsafe reasoning, and disallowed content. I applied harm taxonomies consistently, even in ambiguous scenarios. • Identified and documented subtle unsafe behaviors. • Contributed high-quality human data for AI safety research. • Collaborated asynchronously with research and safety teams. • Documented reasoning to enhance evaluation frameworks.

Not specified

Education

Work History