Newton Mwiti

Key Skills

Software

Clickworker

CrowdSource

Top Subject Matter

No subject matter listed

Top Data Types

Audio

Top Label Types

Cuboid

Freelancer Overview

I am passionate about building reliable AI systems through high-quality data preparation and annotation. My academic and project experience spans NLP and computer vision, including developing sentiment analysis models for social media and image classification tools for inventory management. I have hands-on expertise in data preprocessing, cleaning, and labeling large datasets, using Python, TensorFlow, PyTorch, and NLTK. My work includes optimizing neural networks, collaborating on AI-powered chatbots, and ensuring data integrity for machine learning pipelines. I am comfortable working with diverse data types and thrive in team-based, fast-paced environments focused on AI training data quality.

Labeling Experience

SENIOR IMAGE DATA ANNOTATOR

CrowdsourceImageText Generation

Led image annotation and text generation tasks using the CrowdSource platform, contributing to large-scale AI training datasets between February 2025 and August 2025. The project focused on generating accurate, context-aware textual descriptions for diverse image datasets to support computer vision and multimodal language model development. Responsibilities included analyzing visual content, creating high-quality captions, refining text outputs for clarity and relevance, and ensuring alignment with predefined annotation guidelines. Handled high-volume image batches while maintaining strict quality benchmarks and turnaround timelines. Performed peer reviews, corrected inconsistencies, and ensured compliance with project standards and data security protocols. Maintained high accuracy scores through regular audits and feedback cycles, contributing to improved model performance in image understanding and text generation tasks.

Led image annotation and text generation tasks using the CrowdSource platform, contributing to large-scale AI training datasets between February 2025 and August 2025. The project focused on generating accurate, context-aware textual descriptions for diverse image datasets to support computer vision and multimodal language model development. Responsibilities included analyzing visual content, creating high-quality captions, refining text outputs for clarity and relevance, and ensuring alignment with predefined annotation guidelines. Handled high-volume image batches while maintaining strict quality benchmarks and turnaround timelines. Performed peer reviews, corrected inconsistencies, and ensured compliance with project standards and data security protocols. Maintained high accuracy scores through regular audits and feedback cycles, contributing to improved model performance in image understanding and text generation tasks.

2025 - 2025

data annotator

ClickworkerAudioCuboid

Worked on an audio data labeling project focused on annotating and categorizing audio segments using cuboid-based labeling techniques within the Clickworker platform for the OneForma project. The scope of the project involved identifying, segmenting, and accurately labeling audio clips according to predefined guidelines to support machine learning model training and improvement. Tasks included detecting specific sound events, marking timestamps, classifying speech and non-speech elements, and ensuring precise annotation consistency across datasets. The project consisted of labeling large batches of audio files under strict quality standards, with regular quality audits and performance evaluations. I followed detailed annotation guidelines, maintained high accuracy, met deadline targets, and ensured data integrity through self-review and adherence to quality control protocols. The work contributed to improving AI model performance in audio recognition and processing tasks.

Worked on an audio data labeling project focused on annotating and categorizing audio segments using cuboid-based labeling techniques within the Clickworker platform for the OneForma project. The scope of the project involved identifying, segmenting, and accurately labeling audio clips according to predefined guidelines to support machine learning model training and improvement. Tasks included detecting specific sound events, marking timestamps, classifying speech and non-speech elements, and ensuring precise annotation consistency across datasets. The project consisted of labeling large batches of audio files under strict quality standards, with regular quality audits and performance evaluations. I followed detailed annotation guidelines, maintained high accuracy, met deadline targets, and ensured data integrity through self-review and adherence to quality control protocols. The work contributed to improving AI model performance in audio recognition and processing tasks.

2024 - 2025

Education

Work History