Abdullah Al Swalmi

Key Skills

Software

Appen

Clickworker

Data Annotation Tech

OneForma

Remotasks

Scale AI

Top Subject Matter

No subject matter listed

Top Data Types

Audio

Computer Code Programming

Text

Top Task Types

Audio Recording

Computer Programming Coding

Prompt Response Writing SFT

RLHF

Text Generation

Freelancer Overview

I have over a year of hands-on experience in data labeling and AI training data, working with leading platforms such as Outlier.ai, Appen, OneForma, DataAnnotation, Clickworker, and others. My work has primarily focused on training large language models (LLMs) in both English and Arabic, including regional dialects like UAE, Saudi, and Egyptian Arabic. I contributed extensively to projects involving data annotation, prompt creation, and review across diverse subjects such as math, Python programming, and general language tasks. My ability to deliver accurate and contextually rich annotations set me apart, especially in technical areas requiring both linguistic and domain-specific knowledge. In addition to text-based tasks, I also trained AI on audio clips using platforms like Alignerr and worked on Arabic-English technical translation projects for OneForma. I’ve consistently ranked among the top contributors on Outlier.ai, particularly in math and coding-related projects. My strengths include attention to detail, a deep understanding of both Arabic and English linguistic nuances, and a strong technical background, which enable me to produce high-quality training data that meets the evolving needs of modern AI systems.

Labeling Experience

Tasker

Data Annotation TechTextRLHF

The Argon project on DataAnnotation was focused on evaluating and improving large language model (LLM) outputs through a structured feedback and annotation process. The project involved comparing and ranking AI-generated responses to user prompts based on criteria such as accuracy, relevance, helpfulness, and clarity. Annotators worked on a wide range of topics, including technical subjects like programming and general knowledge tasks in English. The scope included thousands of prompt-response pairs, and the primary task was to rank completions, identify preferred responses, or flag harmful or biased content, helping refine the model’s understanding and alignment. I contributed to the Argon project by providing high-quality, consistent annotations and critical evaluations that directly influenced LLM training. The project emphasized quality assurance through guidelines, calibration tasks, and review feedback, ensuring that only accurate and reliable rankings were accepted.

The Argon project on DataAnnotation was focused on evaluating and improving large language model (LLM) outputs through a structured feedback and annotation process. The project involved comparing and ranking AI-generated responses to user prompts based on criteria such as accuracy, relevance, helpfulness, and clarity. Annotators worked on a wide range of topics, including technical subjects like programming and general knowledge tasks in English. The scope included thousands of prompt-response pairs, and the primary task was to rank completions, identify preferred responses, or flag harmful or biased content, helping refine the model’s understanding and alignment. I contributed to the Argon project by providing high-quality, consistent annotations and critical evaluations that directly influenced LLM training. The project emphasized quality assurance through guidelines, calibration tasks, and review feedback, ensuring that only accurate and reliable rankings were accepted.

2024

Kepler RLHF Reviewer

Scale AITextRLHF

The Kepler RLHF project is a large-scale human-in-the-loop initiative designed to improve the performance and alignment of large language models through reinforcement learning from human feedback. The project involved evaluating and ranking multiple AI-generated responses to prompts across diverse domains, including mathematics, Python programming, general topics, and conversational tasks in both English and Arabic, with emphasis on dialects like UAE, Saudi, and Egyptian Arabic. The scope of the project included thousands of prompts, each with 3 to 5 completions, and labelers were responsible for ranking responses based on their correctness, clarity, coherence, and helpfulness without editing the content. The labeled data was used to train reward models that guide language models to produce more human-aligned outputs. I consistently ranked among the top contributors, especially in math and coding tasks, due to strong analytical skills and linguistic expertise.

The Kepler RLHF project is a large-scale human-in-the-loop initiative designed to improve the performance and alignment of large language models through reinforcement learning from human feedback. The project involved evaluating and ranking multiple AI-generated responses to prompts across diverse domains, including mathematics, Python programming, general topics, and conversational tasks in both English and Arabic, with emphasis on dialects like UAE, Saudi, and Egyptian Arabic. The scope of the project included thousands of prompts, each with 3 to 5 completions, and labelers were responsible for ranking responses based on their correctness, clarity, coherence, and helpfulness without editing the content. The labeled data was used to train reward models that guide language models to produce more human-aligned outputs. I consistently ranked among the top contributors, especially in math and coding tasks, due to strong analytical skills and linguistic expertise.

2024

Education

Work History