Joy F.

Key Skills

Software

Other

Top Subject Matter

Computer vision video action/object annotation for AI training datasets (egocentric/action-recognition)

Quality assurance and dataset validation for computer vision video labeling pipelines

Legal Services & Contract Review

Top Data Types

Video

Image

Text

Document

Top Task Types

Action Recognition

Segmentation

Evaluation/Rating

Text Summarization

Question Answering

Object Detection

Classification

Freelancer Overview

Data Annotator — AI Training Dataset Project, NaijaActions Projects. Brings 4+ years of professional experience across complex professional workflows, research, and quality-focused execution. Core strengths include Internal, Proprietary Tooling, and Other. Education includes Bachelor of Arts (Honours), University of Uyo (2021). AI-training focus includes data types such as Video and labeling workflows including Action Recognition, Evaluation, and Rating.

Labeling Experience

Data Annotator — AI Training Dataset Project, NaijaActions Projects

VideoAction Recognition

Created time-aligned video annotations by segmenting footage using precise timestamps and action boundaries. Labeled primary objects, secondary objects, and tools at macro and micro levels while generating 3,000+ structured segments across 20+ task types. Produced natural language descriptions to support multimodal (vision + language) AI training workflows. • Utilised a proprietary annotation platform for timestamp segmentation and structured output • Interpreted actions, object interactions, and contextual events within video streams • Resolved escalated cases involving overlapping actions, unclear visibility, and multi-object interactions • Produced multimodal-ready datasets combining structured labels and language annotations

2026 - 2026

Quality Assurance Reviewer — Lanfrica / KANAC

OtherVideo

Performed multi-stage quality assurance reviews of annotated video outputs to ensure compliance with project guidelines, taxonomy, and labeling standards. Verified object detection accuracy by confirming label-to-visibility alignment and assessed annotation consistency and completeness across contributors. Validated 10,000+ annotated segments by detecting and correcting missing labels, misclassifications, incorrect mappings, and unusable visual quality. • Evaluated video usability using brightness, sharpness, resolution, and visibility criteria • Detected blur, low resolution, and occluded objects to prevent low-quality approvals • Rejected unreliable data to maintain dataset integrity for AI training • Corrected errors related to missing labels, wrong object mapping, and category mismatches

2025 - 2026

Education

Work History