Hala Chellik

Key Skills

Software

Clickworker

Data Annotation Tech

Remotasks

Toloka

Other

Scale AI

Internal/Proprietary Tooling

Label Studio

CVAT

Top Subject Matter

AI Response Review and Evaluation

LLM Mathematics Problem Solving and Programming Code Evaluation

Record audio in the local language

Top Data Types

Audio

Image

Video

Top Task Types

Audio Recording

Classification

Data Collection

Fine Tuning

Text Generation

Freelancer Overview

I am an experienced Data Engineer and Scientist specializing in AI training data, with expertise in labeling for NLP and CV applications. I have worked on retraining LLMs for mathematical problem-solving, developing code evaluation tools, and creating high-quality datasets. Skilled in data preprocessing, annotation, and visualization, I leverage tools like Selenium and D3.js to deliver precise, impactful results.

Labeling Experience

Audio Recording Annotation in Maghrebi Arabic for Sigma IA

Internal Proprietary ToolingAudioAudio Recording

This project involved labeling and transcribing audio recordings in Maghrebi Arabic, focusing on various accents and speech patterns from the Maghreb region. The task required annotators to accurately transcribe spoken content, including any regional dialects and variations in pronunciation. Additional tasks included categorizing emotions in speech (e.g., happy, sad, angry) and segmenting speech into phrases or sentences. The project aimed to improve speech recognition systems for Maghrebi Arabic and develop more accurate AI models for speech-to-text conversion and emotion detection in this specific dialect. The project included several thousand hours of audio, with a strict quality assurance process to ensure high accuracy.

This project involved labeling and transcribing audio recordings in Maghrebi Arabic, focusing on various accents and speech patterns from the Maghreb region. The task required annotators to accurately transcribe spoken content, including any regional dialects and variations in pronunciation. Additional tasks included categorizing emotions in speech (e.g., happy, sad, angry) and segmenting speech into phrases or sentences. The project aimed to improve speech recognition systems for Maghrebi Arabic and develop more accurate AI models for speech-to-text conversion and emotion detection in this specific dialect. The project included several thousand hours of audio, with a strict quality assurance process to ensure high accuracy.

2024

RWS Multilingual QA

OtherTextFine Tuning

The Llama4 Multilingual Preference Ranking QA project focuses on evaluating the quality of answers generated by language models based on multiple criteria such as relevance, fluency, and linguistic accuracy. Annotators are tasked with ranking the answers and determining if they are appropriate, identifying issues like incorrect classifications or language errors. The goal is to enhance the models' ability to respond correctly and naturally to questions in multiple languages.

The Llama4 Multilingual Preference Ranking QA project focuses on evaluating the quality of answers generated by language models based on multiple criteria such as relevance, fluency, and linguistic accuracy. Annotators are tasked with ranking the answers and determining if they are appropriate, identifying issues like incorrect classifications or language errors. The goal is to enhance the models' ability to respond correctly and naturally to questions in multiple languages.

2024

French audio recordings

Label StudioAudio

French audio recordings

French audio recordings

2024 - 2024

Math Expert Sought for AI Training (French)

Scale AITextPoint Key Point

Remotasks is seeking math experts to help train AI models by providing human feedback. This freelance, remote opportunity involves assessing and evaluating the factuality and relevance of math-related content produced by AI models. Experts will craft and answer math questions, and evaluate AI-generated responses in terms of accuracy and domain relevance. The role requires strong math knowledge, the ability to communicate complex concepts clearly in French, and experience working in the math field. The position offers flexible hours and pays an average of USD $25 per hour, with potential for higher rates based on expertise.

Remotasks is seeking math experts to help train AI models by providing human feedback. This freelance, remote opportunity involves assessing and evaluating the factuality and relevance of math-related content produced by AI models. Experts will craft and answer math questions, and evaluate AI-generated responses in terms of accuracy and domain relevance. The role requires strong math knowledge, the ability to communicate complex concepts clearly in French, and experience working in the math field. The position offers flexible hours and pays an average of USD $25 per hour, with potential for higher rates based on expertise.

2024 - 2024

Data annotation

CVATImageObject Detection

The project is based on identifying the areas where football is located inside the stadium in many pictures.

The project is based on identifying the areas where football is located inside the stadium in many pictures.

2022 - 2022

Show more

Education

Work History