Abdrehim S.

Key Skills

Software

Prodigy

Label Studio

Doccano

Top Subject Matter

Artificial Intelligence & Machine Learning

Finance – Risk Analysis & Fraud Detection

Top Data Types

Image

Text

Audio

Top Task Types

Text Summarization

Object Detection

Classification

Question Answering

Text Generation

Entity (NER) Classification

Freelancer Overview

I have hands‑on experience in AI training and data annotation, gained through both professional roles and intensive programs. At 10 Academy, I worked on financial market analysis and fraud detection pipelines, where I fine‑tuned transformer models, engineered domain‑specific features, and annotated large datasets to improve accuracy. My work raised sentiment model performance from an F1 score of 0.78 to 0.92 and reduced false positives in fraud detection to just 3%. I also built scalable data collection pipelines, tripling throughput by designing asynchronous scraping workflows with retry and backoff strategies. Beyond model development, I have contributed to structured data labeling and evaluation tasks, applying clear annotation guidelines and documenting decisions for reproducibility. I have experience reviewing model outputs, flagging inconsistencies, and providing detailed reasoning to support annotation choices. This combination of technical depth, structured annotation, and clear communication equips me to deliver high‑quality labeled datasets that strengthen AI systems and ensure reliable performance.

Labeling Experience

AI Training & Data Annotation – Sentiment & Fraud Detection

TextClassification

Worked on applied AI projects at 10 Academy’s AI Intensive Program, focusing on sentiment analysis and fraud detection. Fine‑tuned transformer models and annotated domain‑specific datasets, improving sentiment model performance from an F1 score of 0.78 to 0.92. Engineered features for fraud detection pipelines that reduced false positives to 3% while maintaining 85% detection accuracy. Designed scalable ETL pipelines and asynchronous data collection workflows, tripling throughput. Applied structured annotation frameworks, documented labeling decisions for reproducibility, and reviewed model outputs to ensure quality and consistency.

2024 - 2024

Education

Work History