Matthew Grantham

Key Skills

Software

AWS SageMaker

Appen

CloudFactory

Data Annotation Tech

Google Cloud Vertex AI

Top Subject Matter

No subject matter listed

Top Data Types

3D Sensor

Computer Code Programming

Image

Top Task Types

Computer Programming Coding

Data Collection

Geocoding

Prompt Response Writing SFT

Text Generation

Freelancer Overview

As a bilingual freelance AI evaluator, academic writer, and developer with over 10 years of experience, I bring a unique combination of linguistic precision, technical know-how, and analytical thinking to data labeling and training large language models. I’ve contributed to AI evaluation projects through platforms such as Outlier.ai, Toloka, and OpenTrain, where I was responsible for assessing AI-generated responses, identifying logical inconsistencies, rewriting prompts, and rating outputs based on fluency, relevance, and fairness. My academic writing background (APA, MLA, Chicago) reinforces my attention to structure, clarity, and nuanced expression in both French and English. Additionally, my experience in web development and content optimization allows me to evaluate not only linguistic but also contextual and UX-aligned output quality, especially for SaaS platforms and tech-focused applications. I'm comfortable working across strict guidelines and evolving task formats, and I have a proven ability to deliver accurate, high-quality judgments under tight deadlines.

Labeling Experience

LLM Evaluation and Prompt Rating for Human-AI Alignment

Aws SagemakerTextGeocodingRLHF

As a freelance AI trainer, I contributed to multiple large-scale language model tuning projects focused on human-aligned AI evaluation. My responsibilities included evaluating AI-generated responses in English and French, performing A/B comparison tasks, rewriting prompts for safety and clarity, and providing structured ratings on tone, helpfulness, logic, and ethical concerns. I also participated in fine-tuning and prompt crafting (SFT), and reviewed system outputs for sensitive content, coherence, and hallucinations. Each task followed strict annotation guidelines and was performed via proprietary web-based interfaces (Outlier, Toloka, dataannotation). I consistently maintained over 95% agreement with gold-standard labels and ensured output quality met alignment goals for safe LLM deployment.

As a freelance AI trainer, I contributed to multiple large-scale language model tuning projects focused on human-aligned AI evaluation. My responsibilities included evaluating AI-generated responses in English and French, performing A/B comparison tasks, rewriting prompts for safety and clarity, and providing structured ratings on tone, helpfulness, logic, and ethical concerns. I also participated in fine-tuning and prompt crafting (SFT), and reviewed system outputs for sensitive content, coherence, and hallucinations. Each task followed strict annotation guidelines and was performed via proprietary web-based interfaces (Outlier, Toloka, dataannotation). I consistently maintained over 95% agreement with gold-standard labels and ensured output quality met alignment goals for safe LLM deployment.

2022

Education

Work History