Mustapha Mubarak

Key Skills

Software

Data Annotation Tech

Img Lab

LabelImg

Label Studio

Remotasks

Internal/Proprietary Tooling

Top Subject Matter

No subject matter listed

Top Data Types

Audio

Geospatial Tiled Imagery

Image

Medical Dicom

Text

Top Label Types

RLHF

Classification

Question Answering

Evaluation Rating

Prompt Response Writing SFT

Freelancer Overview

I am a data scientist and AI training expert with over six years of experience specializing in data labeling, annotation, and the creation of high-quality datasets for machine learning in healthcare, finance, and STEM domains. My work includes evaluating AI-generated outputs for logical consistency, factual accuracy, and alignment with task guidelines, as well as developing and validating datasets for AI model training and RLHF projects. I am skilled in Python, SQL, and data visualization tools, and have hands-on experience with prompt engineering, response evaluation, and anomaly detection. My background in biomedical engineering and product design allows me to bridge technical and user-focused perspectives, while my strong technical writing ensures clear documentation and effective communication with stakeholders. I am passionate about improving model accuracy and reliability through meticulous data curation, detailed error analysis, and continuous quality assurance.

Labeling Experience

AI Response Evaluation & RLHF Dataset Creation for Financial Fraud Detection

Internal Proprietary ToolingTextQuestion AnsweringRLHF

Served as a subject matter expert and data labeling specialist on a contract AI training project focused on financial anomaly detection and fraud prevention. Responsibilities included: - Evaluating and ranking AI-generated responses using rubric-based frameworks, assessing outputs for accuracy, reasoning quality, coherence, and instruction-following - Writing structured preference justifications explaining ranking decisions to support downstream RLHF fine-tuning of large language models - Authoring complex domain-specific prompts and crafting high-quality reference responses to challenge and improve model reasoning in billing and fraud detection scenarios - Flagging edge cases, hallucinations, and systematic failure patterns across annotation batches, contributing actionable feedback to guideline refinement

Served as a subject matter expert and data labeling specialist on a contract AI training project focused on financial anomaly detection and fraud prevention. Responsibilities included: - Evaluating and ranking AI-generated responses using rubric-based frameworks, assessing outputs for accuracy, reasoning quality, coherence, and instruction-following - Writing structured preference justifications explaining ranking decisions to support downstream RLHF fine-tuning of large language models - Authoring complex domain-specific prompts and crafting high-quality reference responses to challenge and improve model reasoning in billing and fraud detection scenarios - Flagging edge cases, hallucinations, and systematic failure patterns across annotation batches, contributing actionable feedback to guideline refinement

2024

Data Labeling & QA Specialist (Contract), OUTLIER AI

Label StudioTextRLHF

Evaluated AI-generated responses for billing and fraud detection tasks through rubric-based assessment frameworks. Authored STEM-domain prompts and created ground-truth reference responses to challenge LLM reasoning, enhancing dataset diversity. Supplied structured, written justifications that supported RLHF fine-tuning and identified systematic model failure patterns across thousands of annotations. • Ranked model outputs by accuracy, reasoning quality, and instruction-following criteria. • Flagged edge cases and contributed actionable feedback for prompt refinement and guidelines. • Ensured delivery of high-quality data annotation supporting downstream AI models. • Used quality assurance methods to improve annotation guideline iteration.

Evaluated AI-generated responses for billing and fraud detection tasks through rubric-based assessment frameworks. Authored STEM-domain prompts and created ground-truth reference responses to challenge LLM reasoning, enhancing dataset diversity. Supplied structured, written justifications that supported RLHF fine-tuning and identified systematic model failure patterns across thousands of annotations. • Ranked model outputs by accuracy, reasoning quality, and instruction-following criteria. • Flagged edge cases and contributed actionable feedback for prompt refinement and guidelines. • Ensured delivery of high-quality data annotation supporting downstream AI models. • Used quality assurance methods to improve annotation guideline iteration.

2024

AI Data Specialist (Contract), RWSTRAIN AI

Label StudioTextClassification

Annotated and tagged structured text and tabular data according to project guidelines, focusing on billing and financial datasets. Performed pairwise comparison and evaluation tasks on AI-generated outputs for LLM improvement. Delivered calibrated feedback on relevance, factual correctness, and response quality to support annotation pipeline efficiency. • Conducted rigorous quality assurance reviews for schema compliance. • Reported annotation patterns and recurring data quality issues. • Maintained high consistency and accuracy across large-volume datasets. • Contributed to improvement of labeling protocols and data delivery standards.

Annotated and tagged structured text and tabular data according to project guidelines, focusing on billing and financial datasets. Performed pairwise comparison and evaluation tasks on AI-generated outputs for LLM improvement. Delivered calibrated feedback on relevance, factual correctness, and response quality to support annotation pipeline efficiency. • Conducted rigorous quality assurance reviews for schema compliance. • Reported annotation patterns and recurring data quality issues. • Maintained high consistency and accuracy across large-volume datasets. • Contributed to improvement of labeling protocols and data delivery standards.

2023 - 2023

AI Data Annotation Specialist (Contract), TELUS DIGITAL

Label StudioTextClassification

Annotated, categorized, and validated text and structured financial data across multiple content types. Maintained meticulous attention to detail when reviewing high-volume annotation batches for accuracy and factual quality. Collected and curated datasets to meet high-quality thresholds for machine learning pipelines. • Supported guideline clarity and cross-annotator agreement through team collaboration. • Collaborated on global annotation milestones and ambiguity resolutions. • Delivered delivery-ready datasets that met downstream ML team standards. • Ensured each datapoint adhered to project-specific annotation requirements.

Annotated, categorized, and validated text and structured financial data across multiple content types. Maintained meticulous attention to detail when reviewing high-volume annotation batches for accuracy and factual quality. Collected and curated datasets to meet high-quality thresholds for machine learning pipelines. • Supported guideline clarity and cross-annotator agreement through team collaboration. • Collaborated on global annotation milestones and ambiguity resolutions. • Delivered delivery-ready datasets that met downstream ML team standards. • Ensured each datapoint adhered to project-specific annotation requirements.

2022 - 2022

Education

Work History