David Kabundi

Key Skills

Software

Appen

CVAT

Mercor

Roboflow

Scale AI

Toloka

Telus

Other

Top Subject Matter

No subject matter listed

Top Data Types

Audio

Computer Code Programming

Image

Text

Video

Top Label Types

Classification

Computer Programming Coding

Entity Ner Classification

Evaluation Rating

Object Detection

Question Answering

Red Teaming

Freelancer Overview

I am an AI data evaluation and LLM training specialist with experience across text, image, and video annotation for NLP, computer vision, and multimodal model development. My work spans prompt classification, adversarial prompt red teaming, rubric-based evaluation, dataset formatting, and model behavior analysis for LLMs. I have also contributed to structured video segmentation for sports action recognition and computer vision dataset preparation, including PPE image annotation for detection and classification workflows using Roboflow. I excel at interpreting complex guidelines, resolving ambiguous cases, conducting QA reviews, and maintaining high data integrity across annotation pipelines. My toolchain includes MCP, Python, Docker, Roboflow, CVAT, and multimodal dataset formats such as CSV and JSON. I have contributed to projects supporting LLM alignment, safety testing, and structured dataset creation, helping improve model reliability, instruction following, and output quality through consistent data evaluation and preparation.

Labeling Experience

Computer Vision Image Annotation Specialist

RoboflowImageBounding BoxPolygon

Perform detailed image annotation for PPE computer vision model training using Roboflow, including bounding box, polygon, and segmentation labeling for object detection and classification tasks. Ensure pixel-level accuracy, consistent application of labeling guidelines, and high-quality dataset preparation across varied PPE scenarios. Conduct evaluation and QA checks to verify annotation consistency, resolve ambiguous edge cases, and maintain dataset integrity. Contribute to model improvement by delivering clean, structured, and production-ready image datasets.

Perform detailed image annotation for PPE computer vision model training using Roboflow, including bounding box, polygon, and segmentation labeling for object detection and classification tasks. Ensure pixel-level accuracy, consistent application of labeling guidelines, and high-quality dataset preparation across varied PPE scenarios. Conduct evaluation and QA checks to verify annotation consistency, resolve ambiguous edge cases, and maintain dataset integrity. Contribute to model improvement by delivering clean, structured, and production-ready image datasets.

2025

AI Data Verification & NLP Annotation Specialist

MercorTextEntity Ner ClassificationClassification

Performed advanced text prompt annotation and rubric-based evaluation for AI model training at Mercor. Labeled and classified text according to structured rubrics, assessed intents, and evaluated model outputs for correctness, coherence, and adherence to instruction guidelines. Applied judgment to resolve ambiguous cases, verify alignment with evaluation criteria, and ensure consistent interpretation across tasks. Executed QA reviews to maintain data quality standards and contributed to high-integrity datasets used for LLM training and refinement.

Performed advanced text prompt annotation and rubric-based evaluation for AI model training at Mercor. Labeled and classified text according to structured rubrics, assessed intents, and evaluated model outputs for correctness, coherence, and adherence to instruction guidelines. Applied judgment to resolve ambiguous cases, verify alignment with evaluation criteria, and ensure consistent interpretation across tasks. Executed QA reviews to maintain data quality standards and contributed to high-integrity datasets used for LLM training and refinement.

2025

Advanced Video Annotation Specialist (Sports Analytics)

OtherVideoSegmentationClassification

Annotated full-length NBA games for high-precision action recognition and temporal segmentation. Identified and timestamped specific player movements such as backsteps, jumps, impacts, and defensive transitions using frame-accurate review of game footage. Performed player tracking, event classification, and movement analysis to support computer vision model training. Maintained strict timing alignment with in-game clocks and delivered consistent, high-quality annotations under tight turnaround times.

Annotated full-length NBA games for high-precision action recognition and temporal segmentation. Identified and timestamped specific player movements such as backsteps, jumps, impacts, and defensive transitions using frame-accurate review of game footage. Performed player tracking, event classification, and movement analysis to support computer vision model training. Maintained strict timing alignment with in-game clocks and delivered consistent, high-quality annotations under tight turnaround times.

2024

Multimodal Computer Vision Annotator

Scale AIImageBounding BoxClassification

Executed high-precision image annotation for large-scale machine learning datasets using bounding boxes, polygons, segmentation masks, and object detection workflows. Ensured spatial accuracy and consistent label interpretation across thousands of samples by following detailed annotation guidelines. Performed evaluation/rating and quality checks to identify errors, correct inconsistencies, and maintain dataset integrity for computer vision model training.

Executed high-precision image annotation for large-scale machine learning datasets using bounding boxes, polygons, segmentation masks, and object detection workflows. Ensured spatial accuracy and consistent label interpretation across thousands of samples by following detailed annotation guidelines. Performed evaluation/rating and quality checks to identify errors, correct inconsistencies, and maintain dataset integrity for computer vision model training.

2023 - 2024

Education

Work History