For employers

Hire this AI Trainer

Sign in or create an account to invite AI Trainers to your job.

LoginCreate account
Invite to Job
A
Ahmed Marjawi

Ahmed Marjawi

QA AI Trainer & Data Annotator

Egypt flagFayoum, Egypt
$6.00/hrIntermediateAppenScale AIToloka

Key Skills

Software

AppenAppen
Scale AIScale AI
TolokaToloka
Other
ClickworkerClickworker
Don't disclose

Top Subject Matter

Speech AI — Multilingual Audio & Conversational Data Collection
Conversational AI — RLHF, Ranking & Model Evaluation
Multimodal AI — Vision SFT & Prompt Generation

Top Data Types

AudioAudio
TextText
VideoVideo

Top Task Types

Audio RecordingAudio Recording
Computer Programming/CodingComputer Programming/Coding
Data CollectionData Collection
Text GenerationText Generation
Text SummarizationText Summarization
ClassificationClassification
RLHFRLHF
TranscriptionTranscription
Evaluation/RatingEvaluation/Rating
Prompt + Response Writing (SFT)Prompt + Response Writing (SFT)
Bounding BoxBounding Box
Entity (NER) ClassificationEntity (NER) Classification
Object DetectionObject Detection
Question AnsweringQuestion Answering
SegmentationSegmentation

Freelancer Overview

I am a freelance AI data contributor fluent in Arabic (Egyptian dialect) and English, with experience across a wide range of data labeling and AI training tasks. Through platforms like Outlier, I have contributed to projects spanning speech data collection, multimodal vision SFT, RLHF evaluation, prompt generation, transcription, and rubric-based quality assessment across text, audio, image, and video data types. What sets me apart is my ability to adapt quickly to complex task guidelines while maintaining high accuracy across simultaneous projects. I bring native Arabic and professional English linguistic expertise, strong attention to detail, and a solid understanding of AI training workflows from data collection and annotation to preference ranking and model evaluation.

IntermediateArabicEnglish

Labeling Experience

Multilingual Voice Data Collection & Conversational AI Training - Silencio Voice AI

AudioData Collection
Recording single-speaker and multi-speaker voice data in Arabic and English through the Silencio AI platform to support speech AI model training. Tasks include scripted speech recordings and natural live conversations. Contributing to large-scale multilingual datasets used by AI labs across 180+ countries for speech recognition and conversational AI development.

Recording single-speaker and multi-speaker voice data in Arabic and English through the Silencio AI platform to support speech AI model training. Tasks include scripted speech recordings and natural live conversations. Contributing to large-scale multilingual datasets used by AI labs across 180+ countries for speech recognition and conversational AI development.

2026 - Present

Audio Transcription with Annotation Tagging - Neevo AI

AudioTranscription
Transcribing audio recordings into accurate text following strict guidelines covering spelling, grammar, repetitions, stutters, punctuation, capitalization, numbers, and foreign words. Applying structured tags for fillers, vocal noise, background noise, unintelligible speech, and audio cut-offs. Reviewing existing transcripts for accuracy and correcting errors to support speech recognition model training.

Transcribing audio recordings into accurate text following strict guidelines covering spelling, grammar, repetitions, stutters, punctuation, capitalization, numbers, and foreign words. Applying structured tags for fillers, vocal noise, background noise, unintelligible speech, and audio cut-offs. Reviewing existing transcripts for accuracy and correcting errors to support speech recognition model training.

2026 - Present

AI Image Style & Prompt Compliance Evaluation - Toloka

ImageEvaluation Rating
Evaluating pairs of AI-generated images to determine which better matches a given style reference and text prompt. Tasks involve assessing candidates across photographic, illustrative, and compositional criteria — including color scheme, lighting, texture, layout, and subject accuracy — while applying priority rules when prompt and style conflict. Maintaining consistent annotation quality across ongoing production tasks.

Evaluating pairs of AI-generated images to determine which better matches a given style reference and text prompt. Tasks involve assessing candidates across photographic, illustrative, and compositional criteria — including color scheme, lighting, texture, layout, and subject accuracy — while applying priority rules when prompt and style conflict. Maintaining consistent annotation quality across ongoing production tasks.

2026 - Present

AI Image Editing Prompt Evaluation & Comparison - Toloka

ImageEvaluation Rating
Evaluating AI-generated images by comparing two text prompts to identify which better describes the editing applied to source images. Tasks involve reviewing up to 10 source images alongside a generated output and selecting the superior prompt based on editing intent, factual accuracy, image references, and logical consistency. Maintaining accuracy above 80% across ongoing production tasks.

Evaluating AI-generated images by comparing two text prompts to identify which better describes the editing applied to source images. Tasks involve reviewing up to 10 source images alongside a generated output and selecting the superior prompt based on editing intent, factual accuracy, image references, and logical consistency. Maintaining accuracy above 80% across ongoing production tasks.

2026 - Present

First-Person Household Task Video Recording for Hand Movement AI Training - Toloka

VideoData Collection
Collecting egocentric (first-person) video data of household tasks to train AI models in hand movement and activity recognition. Tasks involve recording 5+ minute continuous footage using a downward-mounted camera across categories including Food Preparation, Cleaning, Laundry, Home Organization, and Repair. Each submission is self-labeled by task category, meets Full HD quality standards, and follows strict privacy

Collecting egocentric (first-person) video data of household tasks to train AI models in hand movement and activity recognition. Tasks involve recording 5+ minute continuous footage using a downward-mounted camera across categories including Food Preparation, Cleaning, Laundry, Home Organization, and Repair. Each submission is self-labeled by task category, meets Full HD quality standards, and follows strict privacy

2026 - Present

Education

A

Al-Azhar University

Bachelor of Science, Computer Science

Bachelor of Science
2020 - 2024

Work History

O

Outlier (Scale AI)

QA AI Trainer & Data Annotator

Remote
2025 - Present
A

Ab Tech

Technical Sales Executive

Al Fayyum
2023 - 2025