Dezmonda Abbitt

Key Skills

Software

Labelbox

Top Subject Matter

No subject matter listed

Top Data Types

Image

Video

Top Label Types

Bounding Box

Polygon

Classification

Tracking

Point Key Point

Freelancer Overview

I am an experienced AI Training Specialist with over three years in data labeling and annotation, supporting projects in computer vision, speech recognition, and NLP. My background includes precise annotation of video, image, and audio datasets using tools like Labelbox, CVAT, Supervisely, and Amazon SageMaker Ground Truth. I have extensive hands-on experience with bounding box, polygon, semantic segmentation, and keypoint labeling, as well as audio transcription and sentiment tagging. I consistently maintain high accuracy standards through rigorous quality assurance and collaborate closely with machine learning engineers to refine guidelines and reduce bias. My technical skills in Python and data management, along with my commitment to optimizing datasets, help drive the development of robust and reliable AI models.

Labeling Experience

Multi-Modal AI Data Annotation Specialist – Computer Vision & Speech Models

LabelboxVideoBounding BoxPoint Key Point

Led multi-modal data annotation projects supporting machine learning models in computer vision and speech recognition. Annotated 250,000+ images and video frames using bounding boxes, polygon segmentation, object tracking, and key point labeling for object detection and action recognition systems. Performed 5,000+ hours of audio transcription, speaker diarization, emotion recognition, and NLP entity tagging for LLM and conversational AI training. Maintained 98%+ annotation accuracy through structured QA audits, guideline compliance, inter-annotator agreement checks, and bias reduction practices. Collaborated closely with ML engineers to refine labeling taxonomies and improve dataset performance.

Led multi-modal data annotation projects supporting machine learning models in computer vision and speech recognition. Annotated 250,000+ images and video frames using bounding boxes, polygon segmentation, object tracking, and key point labeling for object detection and action recognition systems. Performed 5,000+ hours of audio transcription, speaker diarization, emotion recognition, and NLP entity tagging for LLM and conversational AI training. Maintained 98%+ annotation accuracy through structured QA audits, guideline compliance, inter-annotator agreement checks, and bias reduction practices. Collaborated closely with ML engineers to refine labeling taxonomies and improve dataset performance.

2024 - 2025

Multi-Modal AI Data Annotation Specialist – Computer Vision & Speech Models

LabelboxImageBounding BoxPolygon

Currently working on a large-scale video annotation project supporting computer vision and action recognition models. Responsible for frame-by-frame labeling of dynamic video datasets, including bounding box annotation, object tracking, polygon segmentation, and key point labeling for human pose estimation. Perform temporal action recognition tagging, identifying specific activities across sequences (e.g., walking, running, object interaction), and ensuring accurate start–end frame classification. Conduct multi-object tracking to maintain consistent ID assignment across frames. Project involves annotating over 120,000+ video frames with strict adherence to detailed labeling guidelines. Maintain 98%+ quality accuracy through internal QA reviews, consistency checks, and inter-annotator validation. Collaborate with machine learning engineers to refine annotation taxonomies, improve edge-case detection, and enhance dataset diversity to reduce bias. Focused on precision, consistency, and

Currently working on a large-scale video annotation project supporting computer vision and action recognition models. Responsible for frame-by-frame labeling of dynamic video datasets, including bounding box annotation, object tracking, polygon segmentation, and key point labeling for human pose estimation. Perform temporal action recognition tagging, identifying specific activities across sequences (e.g., walking, running, object interaction), and ensuring accurate start–end frame classification. Conduct multi-object tracking to maintain consistent ID assignment across frames. Project involves annotating over 120,000+ video frames with strict adherence to detailed labeling guidelines. Maintain 98%+ quality accuracy through internal QA reviews, consistency checks, and inter-annotator validation. Collaborate with machine learning engineers to refine annotation taxonomies, improve edge-case detection, and enhance dataset diversity to reduce bias. Focused on precision, consistency, and

2022 - 2024

Education

Work History