Nevin Okello

Key Skills

Software

Appen

CloudFactory

CrowdSource

CVAT

Hasty

Labelbox

Remotasks

Sama

Telus

VoTT

Scale AI

Top Subject Matter

No subject matter listed

Top Data Types

3D Sensor

Image

Video

Top Task Types

Bounding Box

Cuboid

Evaluation/Rating

Point/Key Point

Segmentation

Freelancer Overview

I am a detail-oriented data annotator with extensive experience in AI training data, specializing in image, text, audio, and video labeling. With a background in Data Science and strong English communication skills, I’ve worked on high-impact projects supporting machine learning model development in industries such as healthcare, finance, and autonomous systems. At CloudFactory and Sama AI, I ensured annotation accuracy by following strict guidelines, utilizing tools like CVAT, Labelbox,Sama, Hasty and Label Studio, and collaborating with cross-functional teams to improve workflows and system performance. What sets me apart is my ability to adapt to evolving project needs, my certification in multiple project-specific workflows, and my hands-on experience handling complex edge cases in computer vision. I bring strong problem-solving abilities, a commitment to quality, and a passion for contributing to innovative AI solutions through precise and reliable data labeling.

Labeling Experience

Vehicles and VRUs

Sama3D SensorBounding BoxPolygon

In a 3D LiDAR point cloud annotation project for autonomous vehicles, the scope involved labeling objects such as vehicles, pedestrians and cyclists in 3D space using cuboids and then generating a linked bouning box of the same object in 2D space. The task required frame-by-frame annotation to ensure temporal consistency and accurate object tracking across scenes. Quality measures included strict adherence to project guidelines, cross-verification of annotations, precision in object boundaries, consistency checks, and regular QA reviews to maintain high accuracy and reliability for model training.

In a 3D LiDAR point cloud annotation project for autonomous vehicles, the scope involved labeling objects such as vehicles, pedestrians and cyclists in 3D space using cuboids and then generating a linked bouning box of the same object in 2D space. The task required frame-by-frame annotation to ensure temporal consistency and accurate object tracking across scenes. Quality measures included strict adherence to project guidelines, cross-verification of annotations, precision in object boundaries, consistency checks, and regular QA reviews to maintain high accuracy and reliability for model training.

2023 - 2025

Ridables Object Detection

VottImageBounding Box

The image annotation project for a ridable like a motorcycle focused on identifying and labeling specific parts such as wheels, headlights, handlebars, engine, seat, and exhaust using bounding boxes and polygon annotations. This supported object recognition, damage detection, and parts inventory systems. The task required high precision to distinguish between similar components across various motorcycle models. Quality measures included detailed labeling guidelines, multi-level QA reviews, accuracy thresholds, and consistency audits to ensure precise, complete, and uniform annotations.

The image annotation project for a ridable like a motorcycle focused on identifying and labeling specific parts such as wheels, headlights, handlebars, engine, seat, and exhaust using bounding boxes and polygon annotations. This supported object recognition, damage detection, and parts inventory systems. The task required high precision to distinguish between similar components across various motorcycle models. Quality measures included detailed labeling guidelines, multi-level QA reviews, accuracy thresholds, and consistency audits to ensure precise, complete, and uniform annotations.

2025

Baseball Location

HastyImageBounding BoxPolygon

The image annotation project for the baseball sports industry focused on labeling player actions, equipment like bat, glove, ball, field zones, and event outcomes (e.g., pitch type, swing, catch) using bounding boxes and keypoint annotation. The data supported player performance analysis, play tracking, and training model development for sports analytics. Tasks required frame-by-frame consistency, especially for fast-paced sequences. Quality measures included double-blind reviews, precision benchmarks, and continuous feedback loops to maintain annotation accuracy, completeness, and temporal consistency.

The image annotation project for the baseball sports industry focused on labeling player actions, equipment like bat, glove, ball, field zones, and event outcomes (e.g., pitch type, swing, catch) using bounding boxes and keypoint annotation. The data supported player performance analysis, play tracking, and training model development for sports analytics. Tasks required frame-by-frame consistency, especially for fast-paced sequences. Quality measures included double-blind reviews, precision benchmarks, and continuous feedback loops to maintain annotation accuracy, completeness, and temporal consistency.

2025

Image Segmentation

CVATImageSegmentation

The image segmentation annotation project for the real estate industry involved labeling property images by segmenting key structural elements such as walls, doors, windows, roofs, and floors. The task required pixel-level precision to enable accurate object detection and classification for virtual staging, property analysis, and automated valuation models. Annotations were performed using specialized tools with clear class definitions and labeling protocols. Quality measures included regular QA audits, inter-annotator agreement checks, and adherence to project guidelines to ensure accuracy, consistency, and completeness of the segmented regions.

The image segmentation annotation project for the real estate industry involved labeling property images by segmenting key structural elements such as walls, doors, windows, roofs, and floors. The task required pixel-level precision to enable accurate object detection and classification for virtual staging, property analysis, and automated valuation models. Annotations were performed using specialized tools with clear class definitions and labeling protocols. Quality measures included regular QA audits, inter-annotator agreement checks, and adherence to project guidelines to ensure accuracy, consistency, and completeness of the segmented regions.

2023

Human Body Fitness

SamaVideoPoint Key Point

The image annotation project for human body fitness involved keypoint skeleton tracking to label major joints like shoulders, elbows, knees, and ankles across workout videos and fitness images. The goal was to capture posture, movement patterns, and alignment for applications in exercise correction, pose estimation, and fitness coaching AI. Annotations were done frame-by-frame to ensure temporal accuracy. Quality measures included strict adherence to anatomical keypoint definitions, use of QA sampling, inter-annotator consistency checks, and validation against reference poses to ensure high accuracy and reliability.

The image annotation project for human body fitness involved keypoint skeleton tracking to label major joints like shoulders, elbows, knees, and ankles across workout videos and fitness images. The goal was to capture posture, movement patterns, and alignment for applications in exercise correction, pose estimation, and fitness coaching AI. Annotations were done frame-by-frame to ensure temporal accuracy. Quality measures included strict adherence to anatomical keypoint definitions, use of QA sampling, inter-annotator consistency checks, and validation against reference poses to ensure high accuracy and reliability.

2021

Education

Work History