Abdulrahman A.

Key Skills

Software

Other

Top Subject Matter

Generative AI training datasets (computer vision)

E-commerce/product vision datasets

Top Data Types

Image

Text

Video

Top Task Types

Data Collection

Bounding Box

Polygon

Segmentation

Point/Key Point

RLHF

Freelancer Overview

Ironic X | Image & Video Annotation Lead / Founder / Lead AI & Brand Designer. Brings 2+ years of professional experience across complex professional workflows, research, and quality-focused execution. Core strengths include Internal and Proprietary Tooling. Education includes Bachelor of Science, Ahmadu Bello University (2025) and Certificate, Niger State Skills for the Future Initiative (2024). AI-training focus includes data types such as Image and Video and labeling workflows including Bounding Box, Polygon, and Segmentation.

Labeling Experience

Ironic X | Image & Video Annotation Lead / Founder / Lead AI & Brand Designer

ImageBounding BoxPolygon

Served as Image & Video Annotation Lead, annotating and quality-reviewing 5,000+ frames for generative AI training datasets. Produced bounding boxes, polygon masks, semantic segmentation labels, and keypoint annotations while maintaining a reported >98% accuracy rate across submitted batches. Delivered RLHF-style human feedback by evaluating prompt alignment, spatial coherence, and visual fidelity to training specifications. • Annotated visual datasets for generative AI model training and sustained high acceptance accuracy. • Performed annotation QA and re-review reduction via internal checklists and rubrics. • Evaluated model outputs using human feedback loops focused on alignment and visual fidelity. • Supported studio-wide workflow improvements to reduce downstream model error rates.

2023 - Present

Arancio Collection | Data Annotator

ImagePolygon

Worked as a Data Annotator for product and lifestyle image datasets, performing polygon masking and structured attribute tagging. Completed category classification and multi-label annotations (e.g., color, material, category) across 3,000+ images using consistent labeling protocols. Ensured dataset integrity by flagging edge cases, ambiguous labels, and corrupted sources that increased downstream training noise. • Annotated 3,000+ product/lifestyle images with polygon masks and attribute tags. • Maintained batch-level consistency across varying lighting, occlusion, and viewpoints. • Applied e-commerce labeling workflows including bounding boxes and background segmentation. • Escalated corrupted files and ambiguous classes to improve dataset quality.

2024 - 2025

Education

Work History