Ahmed Abdelwahed

Key Skills

Software

Appen

Clickworker

Remotasks

Scale AI

Top Subject Matter

No subject matter listed

Top Data Types

Audio

Text

Top Task Types

Audio Recording

Evaluation Rating

Prompt Response Writing SFT

Text Generation

Translation Localization

Freelancer Overview

I bring a unique blend of expertise in AI training data, content evaluation, and prompt engineering, grounded in real-world experience across marketing, education, and linguistics. I have worked on advanced projects such as cross-lingual semantic textual similarity with register and stylistic considerations (XSTS+R+P), where I conducted sentence-level evaluations of Arabic-English translations for semantic fidelity and stylistic equivalence. My background also includes assessing large language model (LLM) outputs for truthfulness, clarity, neutrality, and contextual accuracy, particularly in time-sensitive or culturally nuanced topics. In addition to AI-specific tasks, I leverage years of experience in social media content strategy and educational content design, especially for Arabic-speaking audiences in Egypt and the Gulf region. This allows me to contribute to data labeling efforts that require cultural awareness, deep language sensitivity, and domain-specific context — whether in marketing, education, or interactive user engagement. My skill set is especially valuable for tasks involving prompt design, content categorization, quality control, and multilingual alignment.

Labeling Experience

Cross-Lingual Semantic Similarity & Prompt Evaluation for LLMs

AppenTextClassificationText Summarization

Worked on multiple high-level AI training projects focused on evaluating and annotating natural language data for large language models. One key project involved Cross-lingual Semantic Textual Similarity with Register and Politeness (XSTS+R+P), where I assessed Arabic-English sentence pairs for semantic equivalence, tone, and stylistic accuracy. Tasks included applying 1–5 rating scales with justification, flagging semantic mismatches, and ensuring register consistency. Additionally, I contributed to LLM output evaluation, focusing on truthfulness, clarity, neutrality, and relevance across diverse domains (e.g., marketing, education, general knowledge). This included writing and refining prompts based on user intent (e.g., open QA, closed QA, generation, brainstorming) and rating model outputs in line with strict quality guidelines. Projects followed rigorous accuracy and consistency metrics, with multi-rater agreement protocols and daily feedback loops.

Worked on multiple high-level AI training projects focused on evaluating and annotating natural language data for large language models. One key project involved Cross-lingual Semantic Textual Similarity with Register and Politeness (XSTS+R+P), where I assessed Arabic-English sentence pairs for semantic equivalence, tone, and stylistic accuracy. Tasks included applying 1–5 rating scales with justification, flagging semantic mismatches, and ensuring register consistency. Additionally, I contributed to LLM output evaluation, focusing on truthfulness, clarity, neutrality, and relevance across diverse domains (e.g., marketing, education, general knowledge). This included writing and refining prompts based on user intent (e.g., open QA, closed QA, generation, brainstorming) and rating model outputs in line with strict quality guidelines. Projects followed rigorous accuracy and consistency metrics, with multi-rater agreement protocols and daily feedback loops.

2024 - 2024

Education

Work History