Mesbah Ryad

Key Skills

Software

Appen

Labelbox

Scale AI

Top Subject Matter

No subject matter listed

Top Data Types

Audio

Image

Text

Top Task Types

Evaluation Rating

Prompt Response Writing SFT

Question Answering

RLHF

Freelancer Overview

Since 2024, I have been working as an AI model trainer on the Outlier platform, specializing in evaluating and refining artificial intelligence models. My role involves crafting prompts, analyzing model responses, and identifying potential weaknesses to enhance their performance. I work in both English and French, ensuring linguistic accuracy and adaptability. This experience has strengthened my expertise in AI behavior, prompt engineering, and critical analysis of machine learning systems.

Labeling Experience

Cypher Evals

Scale AITextEvaluation Rating

I worked on the Cypher Evals project at Outlier, where I evaluated and stress-tested advanced language-model behavior across a wide range of tasks. My work involved assessing model outputs for correctness, reasoning quality, creativity, safety, and adherence to instructions. I performed detailed annotations, identified reasoning flaws, and graded multi-step solutions to help improve the model’s reliability

I worked on the Cypher Evals project at Outlier, where I evaluated and stress-tested advanced language-model behavior across a wide range of tasks. My work involved assessing model outputs for correctness, reasoning quality, creativity, safety, and adherence to instructions. I performed detailed annotations, identified reasoning flaws, and graded multi-step solutions to help improve the model’s reliability

2024

Grassland

Scale AIAudioAudio Recording

I worked on the Grassland Audio project at Outlier, where I evaluated and refined AI models designed to understand, classify, and respond to audio inputs. My tasks included reviewing audio clips, analyzing model transcriptions and interpretations, and assessing accuracy, clarity, tone, and contextual understanding. I provided structured feedback on errors such as misheard phrases, background noise confusion, speaker misidentification, or incorrect intent extraction.

I worked on the Grassland Audio project at Outlier, where I evaluated and refined AI models designed to understand, classify, and respond to audio inputs. My tasks included reviewing audio clips, analyzing model transcriptions and interpretations, and assessing accuracy, clarity, tone, and contextual understanding. I provided structured feedback on errors such as misheard phrases, background noise confusion, speaker misidentification, or incorrect intent extraction.

2025 - 2025

Pangolin Safety

Scale AIImagePrompt Response Writing SFT

I worked on the Pangolin Safety Image project at Outlier, focusing on evaluating and improving the model’s ability to understand, classify, and respond safely to visual content. My responsibilities included reviewing AI-generated image interpretations, identifying safety risks, and ensuring alignment with content policies. I analyzed images for potential violations—such as sensitive, harmful, or restricted content—and provided structured feedback to help refine the model’s detection accuracy and reduce false positives/negatives.

I worked on the Pangolin Safety Image project at Outlier, focusing on evaluating and improving the model’s ability to understand, classify, and respond safely to visual content. My responsibilities included reviewing AI-generated image interpretations, identifying safety risks, and ensuring alignment with content policies. I analyzed images for potential violations—such as sensitive, harmful, or restricted content—and provided structured feedback to help refine the model’s detection accuracy and reduce false positives/negatives.

2025 - 2025

Cypher RLHF

Scale AITextRLHF

I worked on the Cypher RLHF project at Outlier, contributing to the development and refinement of advanced language-model reasoning. My role involved evaluating model outputs, analyzing multi-step reasoning chains, and providing high-quality human feedback to guide model alignment. I focused on improving clarity, logical consistency, factual accuracy, and safety across generated responses

I worked on the Cypher RLHF project at Outlier, contributing to the development and refinement of advanced language-model reasoning. My role involved evaluating model outputs, analyzing multi-step reasoning chains, and providing high-quality human feedback to guide model alignment. I focused on improving clarity, logical consistency, factual accuracy, and safety across generated responses

2024 - 2025

Education

Work History