David Mwangi

Key Skills

Software

Appen

Clickworker

Dataturk

Mindrift

OneForma

Remotasks

Snorkel AI

Toloka

Telus

Scale AI

Top Subject Matter

No subject matter listed

Top Data Types

Geospatial Tiled Imagery

Text

Video

Top Task Types

Classification

Diagnosis

Evaluation Rating

Prompt Response Writing SFT

RLHF

Freelancer Overview

With a strong foundation in mathematics and a deep commitment to detail, I bring hands-on experience in AI training data projects across multiple domains. I have contributed to data labeling and annotation tasks for leading platforms such as Mindrift, Telus AI, OneForma, and Toloka. My work spans e-commerce product image and text labeling, where I helped structure large-scale datasets to improve product search and categorization; social media content annotation, supporting safety and moderation algorithms; and English LLM evaluation, where I reviewed AI-generated outputs for factual accuracy, coherence, and ethical alignment. What sets me apart is my ability to combine analytical thinking with linguistic precision. My academic background at the PhD level honed my attention to logic and structure, which I now apply to complex annotation tasks. Whether evaluating the tone of a chatbot response or tagging object boundaries in images, I approach every task with care, consistency, and a deep understanding of how high-quality data shapes smarter AI. I’m equally comfortable following strict guidelines and providing insightful feedback that helps refine annotation protocols.

Labeling Experience

E-commerce Image & Text Labeling for Product Categorization

ClickworkerImageBounding BoxClassification

In this project, I annotated and labeled thousands of e-commerce product images and titles to improve machine learning models used for product recommendation, visual search, and categorization. My tasks included drawing bounding boxes around key visual features (e.g., logos, packaging), assigning accurate product categories and attributes, and verifying text metadata like titles and specifications. The project involved high-volume labeling with strict accuracy standards (>95%), reviewed through multiple QA rounds. I ensured compliance with dynamic guidelines, maintained label consistency, and met tight deadlines while delivering high-quality work that directly contributed to AI training for global retail clients.

In this project, I annotated and labeled thousands of e-commerce product images and titles to improve machine learning models used for product recommendation, visual search, and categorization. My tasks included drawing bounding boxes around key visual features (e.g., logos, packaging), assigning accurate product categories and attributes, and verifying text metadata like titles and specifications. The project involved high-volume labeling with strict accuracy standards (>95%), reviewed through multiple QA rounds. I ensured compliance with dynamic guidelines, maintained label consistency, and met tight deadlines while delivering high-quality work that directly contributed to AI training for global retail clients.

2021 - 2024

Education

Work History