For employers

Hire this AI Trainer

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

LoginCreate account
Invite to Job
Abdul Ibrahim

Abdul Ibrahim

Expert in STEM prompt curation & AI data labeling for LLMs and CV models

USA flagSan Diego, Usa
$25.00/hrExpertCVATLabelboxScale AI

Key Skills

Software

CVATCVAT
LabelboxLabelbox
Scale AIScale AI
SuperAnnotateSuperAnnotate
Other

Top Subject Matter

No subject matter listed

Top Data Types

Computer Code ProgrammingComputer Code Programming
ImageImage
TextText

Top Task Types

Bounding Box
Computer Programming Coding
Object Detection
Text Summarization

Freelancer Overview

I’m a seasoned AI training specialist with 5+ years of experience in data labeling and evaluation for large language models and computer vision systems. I’ve contributed to projects for OpenAI, Google, and Meta, specializing in STEM-focused prompt creation, annotation, and response validation. My work includes identifying model hallucinations, correcting reasoning errors, and curating high-quality datasets in math, physics, and Python. I've also led labeling tasks in computer vision, applying bounding boxes, QA tagging, and metadata labeling using tools like Labelbox, CVAT, and OpenCV. My real-time object detection projects required precise annotation and calibration of training data to support robust performance. I bring strong attention to detail, a deep technical foundation, and proven success working in fast-paced, fully remote environments.

ExpertArabicEnglish

Labeling Experience

Financial Time-Series Labeling for Stock Volatility Prediction

OtherTextClassification
Labeled and structured financial time-series data to train a model predicting stock volatility patterns. Defined volatility classes using statistical thresholds, tagged anomalous events, and curated clean input features for supervised learning. Processed hundreds of daily stock records using Python, applying rule-based and statistical labeling logic. Maintained high data integrity with validation scripts and cross-checking for mislabeled outliers.

Labeled and structured financial time-series data to train a model predicting stock volatility patterns. Defined volatility classes using statistical thresholds, tagged anomalous events, and curated clean input features for supervised learning. Processed hundreds of daily stock records using Python, applying rule-based and statistical labeling logic. Maintained high data integrity with validation scripts and cross-checking for mislabeled outliers.

2024 - 2024

Ground Truth Validation for ML Classifier Benchmarking

OtherTextDiagnosis
Evaluated the performance of ML classifiers on UCI datasets by validating target labels, identifying inconsistencies, and cleaning input features. Tasks included verifying class distributions, checking mislabeled entries, and preparing datasets for cross-validation. Ensured labeling accuracy and dataset integrity to enable fair model comparison across Random Forest, SVM, and Decision Tree algorithms. Maintained reproducibility with documented preprocessing steps and quality controls.

Evaluated the performance of ML classifiers on UCI datasets by validating target labels, identifying inconsistencies, and cleaning input features. Tasks included verifying class distributions, checking mislabeled entries, and preparing datasets for cross-validation. Ensured labeling accuracy and dataset integrity to enable fair model comparison across Random Forest, SVM, and Decision Tree algorithms. Maintained reproducibility with documented preprocessing steps and quality controls.

2023 - 2023

Real-Time Object Detection for Auto Aim in Games

OtherImageBounding BoxObject Detection
Labeled over 10,000 in game frames to train a YOLOv5 model for enemy detection in Unreal Tournament 1999. Tasks included drawing bounding boxes around enemy players, enemy body parts, classifying object types, and cleaning mislabeled data. Ensured frame to frame consistency and precise hitbox alignment for high model accuracy. Quality control included manual review, overlap checks, and pixel accuracy thresholds.

Labeled over 10,000 in game frames to train a YOLOv5 model for enemy detection in Unreal Tournament 1999. Tasks included drawing bounding boxes around enemy players, enemy body parts, classifying object types, and cleaning mislabeled data. Ensured frame to frame consistency and precise hitbox alignment for high model accuracy. Quality control included manual review, overlap checks, and pixel accuracy thresholds.

2023 - 2023

Education

U

University of California, San Diego

Bachelor of Science, Computational Neuroscience

Bachelor of Science
2020 - 2024

Work History

S

Scale AI

AI Trainer Specialist | STEM & Python Expert for LLMs

San Diego
2024 - Present
G

Grumpy Chef

eCommerce Manager

San Diego
2020 - 2024