Ahmed Hosny

Key Skills

Software

Clickworker

Data Annotation Tech

Labelbox

OneForma

Scale AI

Toloka

Telus

Internal/Proprietary Tooling

Other

Top Subject Matter

No subject matter listed

Top Data Types

Audio

Image

Text

Top Task Types

Classification

Evaluation Rating

Fine Tuning

Prompt Response Writing SFT

RLHF

Freelancer Overview

Sure, here’s a professional and concise summary tailored to your experience in data labeling and AI training: --- I have extensive experience in AI training and data labeling, with a strong focus on Arabic language tasks. My work includes evaluating large language models (LLMs), rewriting prompts and responses, annotating text for instruction tuning, and rating AI-generated outputs based on quality, helpfulness, accuracy, and tone. I’ve contributed to diverse projects involving natural conversation, storytelling, question answering, and classification tasks, helping improve model performance across Arabic and English. What sets me apart is my deep understanding of linguistic nuances in Arabic, my ability to follow strict labeling guidelines, and my experience in multiple domains such as Health, Religion, and General Knowledge. I’m skilled in ensuring data quality, maintaining cultural sensitivity, and optimizing training datasets for multilingual AI systems.

Labeling Experience

Multimodal AI Specialist – Video & Visual Data Classification

OtherVideoClassification

This project focused on the classification of video and visual data to support multimodal large language model (LLM) development. My role involved reviewing video content and annotating it based on predefined classification criteria such as scene type, action categories, sentiment, and relevance to associated textual prompts. Using a proprietary labeling platform, I handled a high volume of data while maintaining strict quality control through regular audits, inter-rater consistency checks, and continuous guideline updates. The labeled data contributed to training and aligning LLMs to better understand and respond to visual inputs.

This project focused on the classification of video and visual data to support multimodal large language model (LLM) development. My role involved reviewing video content and annotating it based on predefined classification criteria such as scene type, action categories, sentiment, and relevance to associated textual prompts. Using a proprietary labeling platform, I handled a high volume of data while maintaining strict quality control through regular audits, inter-rater consistency checks, and continuous guideline updates. The labeled data contributed to training and aligning LLMs to better understand and respond to visual inputs.

2024

AI Data Trainer

Data Annotation TechTextPrompt Response Writing SFT

As an AI Data Trainer for LLM development, I worked on a large-scale supervised fine-tuning (SFT) project involving prompt and response writing. The scope of the project included crafting high-quality, human-like responses to a wide variety of text prompts across multiple domains, with a focus on accuracy, coherence, and alignment with ethical and safety guidelines. Using the Data Annotation Tech platform, I generated and reviewed thousands of prompt-response pairs to train large language models. Quality measures included adherence to detailed annotation guidelines, peer review, regular feedback cycles, and continuous calibration to maintain consistency and excellence across the dataset.

As an AI Data Trainer for LLM development, I worked on a large-scale supervised fine-tuning (SFT) project involving prompt and response writing. The scope of the project included crafting high-quality, human-like responses to a wide variety of text prompts across multiple domains, with a focus on accuracy, coherence, and alignment with ethical and safety guidelines. Using the Data Annotation Tech platform, I generated and reviewed thousands of prompt-response pairs to train large language models. Quality measures included adherence to detailed annotation guidelines, peer review, regular feedback cycles, and continuous calibration to maintain consistency and excellence across the dataset.

2024

LLM Localization

Scale AIImageEvaluation Rating

This project involved evaluating and rating image-based data used for LLM localization tasks. Using the Scale AI platform, I assessed the relevance, accuracy, and contextual fit of visual inputs intended to align with large language models. The task required meticulous attention to detail in judging whether images correctly corresponded to associated prompts, concepts, or cultural references across various locales. The project covered tens of thousands of image samples, with strict adherence to quality standards through regular calibration sessions, benchmark reviews, and feedback-driven iterations to ensure consistent and high-quality evaluations.

This project involved evaluating and rating image-based data used for LLM localization tasks. Using the Scale AI platform, I assessed the relevance, accuracy, and contextual fit of visual inputs intended to align with large language models. The task required meticulous attention to detail in judging whether images correctly corresponded to associated prompts, concepts, or cultural references across various locales. The project covered tens of thousands of image samples, with strict adherence to quality standards through regular calibration sessions, benchmark reviews, and feedback-driven iterations to ensure consistent and high-quality evaluations.

2024 - 2024

Education

Work History