Mohamed Taha Ait El Houcine

Key Skills

Software

Other

Appen

Top Subject Matter

No subject matter listed

Top Data Types

Image

Audio

Top Label Types

Object Detection

Data Collection

Prompt Response Writing SFT

Audio Recording

Transcription

Translation Localization

Freelancer Overview

I am a results-driven data quality and labeling specialist with hands-on experience in AI training data, content moderation, and e-commerce operations. My background includes maintaining over 90% QA accuracy in large-scale labeling projects for TikTok, leading calibration sessions, and collaborating with policy owners and BPO partners to ensure precise policy enforcement and market alignment. I have a strong track record in training and onboarding new QA specialists, conducting root cause analysis, and driving continuous process improvements. My skill set spans data analysis, Excel (pivot tables, VLOOKUP, data visualization), user behavior analysis, and reporting actionable insights to optimize project outcomes. I thrive in fast-paced, cross-functional environments and am passionate about leveraging data-driven solutions to enhance operational efficiency and the quality of AI training data, especially in domains such as e-commerce, search, and content safety. Fluent in English, French, Arabic, and German, I am adept at adapting to diverse markets and collaborating with global teams to deliver measurable results.

Labeling Experience

Outlier AETHER Generalist

OtherImageObject DetectionData Collection

This project is an ongoing global AI training initiative focused on improving Large Language Models (LLMs) and generative AI systems through diverse human feedback tasks. The primary objective is to align AI behavior with human preferences, making models more accurate, helpful, and aligned with user intent. Tasks are short, flexible, and varied, requiring attention to detail and human judgment. They include: Comparing and evaluating AI-generated text and images for accuracy, quality, and nuance. Spotting differences and subtle details between visuals. Writing short dialogues and natural conversations based on prompts or clips. Describing visuals clearly and concisely. Adopting a "user mindset" to generate natural data and follow specific rubrics to address user needs in a structured way. Quality measures emphasize care, consistency, and attention to detail, as accuracy in tasks unlocks more opportunities.

This project is an ongoing global AI training initiative focused on improving Large Language Models (LLMs) and generative AI systems through diverse human feedback tasks. The primary objective is to align AI behavior with human preferences, making models more accurate, helpful, and aligned with user intent. Tasks are short, flexible, and varied, requiring attention to detail and human judgment. They include: Comparing and evaluating AI-generated text and images for accuracy, quality, and nuance. Spotting differences and subtle details between visuals. Writing short dialogues and natural conversations based on prompts or clips. Describing visuals clearly and concisely. Adopting a "user mindset" to generate natural data and follow specific rubrics to address user needs in a structured way. Quality measures emphasize care, consistency, and attention to detail, as accuracy in tasks unlocks more opportunities.

2025

Appen transcriber/translator

AppenAudioTranslation LocalizationTranscription

This project involves data labeling for the development and improvement of Large Language Models (LLMs) and AI-powered speech recognition and translation systems. The core task is to process audio data, focusing on translation from Arabic to English and from French to English, as well as transcription of audio content in the source languages. Specific tasks include: Transcribing diverse Arabic and French audio materials into text, adhering to specific linguistic guidelines and annotation standards. Translating and localizing the content to reflect cultural and dialectical variations, ensuring linguistic accuracy and natural fluency in English. Reviewing and quality checking existing transcriptions and translations to ensure a high level of fidelity, consistency, and alignment with project rubrics. The project aims to provide high-quality, human-annotated datasets to train AI models to understand diverse linguistic and cultural nuances effectively.

This project involves data labeling for the development and improvement of Large Language Models (LLMs) and AI-powered speech recognition and translation systems. The core task is to process audio data, focusing on translation from Arabic to English and from French to English, as well as transcription of audio content in the source languages. Specific tasks include: Transcribing diverse Arabic and French audio materials into text, adhering to specific linguistic guidelines and annotation standards. Translating and localizing the content to reflect cultural and dialectical variations, ensuring linguistic accuracy and natural fluency in English. Reviewing and quality checking existing transcriptions and translations to ensure a high level of fidelity, consistency, and alignment with project rubrics. The project aims to provide high-quality, human-annotated datasets to train AI models to understand diverse linguistic and cultural nuances effectively.

2018 - 2019

Education

Work History