Ntando Madonsela

Key Skills

Software

Clickworker

CrowdSource

Data Annotation Tech

Label Studio

Mindrift

OneForma

Toloka

Internal/Proprietary Tooling

Top Subject Matter

No subject matter listed

Top Data Types

Image

Text

Video

Top Task Types

Action Recognition

Classification

Data Collection

Emotion Recognition

Evaluation/Rating

Freelancer Overview

I have extensive experience in data labeling and AI training, having worked as a data annotator for Toloka AI and SuperAnnotate. My work has involved labeling and verifying high-quality datasets for computer vision, natural language processing, and multimodal AI systems. I have contributed to projects supporting autonomous driving, object detection, image segmentation, and text classification, ensuring accuracy and consistency across complex datasets. My expertise includes using advanced annotation tools such as Labelbox, CVAT, Supervisely, and Roboflow, as well as performing quality assurance and dataset validation for AI model training. I am skilled in identifying data inconsistencies, optimizing labeling workflows, and applying precise labeling standards that improve model performance. My attention to detail, understanding of AI data requirements, and ability to adapt across diverse domains set me apart as a reliable and efficient contributor to AI development projects.

Labeling Experience

Data Annotator

TolokaTextText GenerationPrompt Response Writing SFT

Contributed to an AI data annotation and model training project focused on improving prompt–response accuracy and contextual understanding in large language models. The project followed the STF (Source–Task–Feedback) framework to ensure data quality and performance optimization. Source: Collected and curated datasets from multilingual text and visual content to build diverse and representative training material. Task: Created, labeled, and evaluated prompt–response pairs for natural language generation, reasoning, and summarization tasks. Annotated data for sentiment analysis, intent detection, and content classification. Feedback: Conducted quality assurance reviews, provided structured feedback on model outputs, and refined prompts to enhance response accuracy, coherence, and tone alignment. This project combined human annotation expertise with iterative AI evaluation to strengthen model reliability and reduce error rates across English and multilingual datasets. Tools used inclu

Contributed to an AI data annotation and model training project focused on improving prompt–response accuracy and contextual understanding in large language models. The project followed the STF (Source–Task–Feedback) framework to ensure data quality and performance optimization. Source: Collected and curated datasets from multilingual text and visual content to build diverse and representative training material. Task: Created, labeled, and evaluated prompt–response pairs for natural language generation, reasoning, and summarization tasks. Annotated data for sentiment analysis, intent detection, and content classification. Feedback: Conducted quality assurance reviews, provided structured feedback on model outputs, and refined prompts to enhance response accuracy, coherence, and tone alignment. This project combined human annotation expertise with iterative AI evaluation to strengthen model reliability and reduce error rates across English and multilingual datasets. Tools used inclu

2025 - 2025

Education

Work History