Obinna Okonkwo

Key Skills

Software

Clickworker

CloudFactory

CVAT

Data Annotation Tech

Deep Systems

Figure Eight

HiveMind

Labelbox

Label Studio

Mercor

OneForma

OpenCV AI Kit (OAK)

Remotasks

SuperAnnotate

V7 Labs

Scale AI

Appen

Top Subject Matter

No subject matter listed

Top Data Types

Audio

Document

Text

Top Label Types

Data Collection

Evaluation Rating

Mapping

Prompt Response Writing SFT

Question Answering

Freelancer Overview

As a meticulous Data Labeling Specialist with over 2 years of experience, I specialize in preparing high-quality training datasets for cutting-edge AI models across computer vision, natural language processing, and LLM evaluation. My work includes annotating images, video, and text data for applications such as autonomous vehicles, sentiment analysis, and conversational AI. I am skilled in using industry tools like Labelbox, Scale AI, and Prodigy, and have a strong eye for detail, ensuring data integrity and consistency for machine learning readiness. I bring a unique edge through my background in mechanical engineering, computational linguistics, and software QA, allowing me to bridge the gap between technical requirements and real-world usability. Having contributed to projects such as fine-tuning multilingual LLMs and improving object detection models for safety-critical systems, I excel in delivering accurate, scalable annotations that drive model performance and reliability.

Labeling Experience

Image Annotation for Retail & E-Commerce Product Recognition

Scale AIImageBounding Box

Executed precise image annotation projects for TELUS Digital to enhance computer vision models for retail and e-commerce applications. Tasks included drawing high-accuracy bounding boxes and polygon segments to identify and isolate thousands of individual products within complex scenes. This involved detailed classification of items based on category, brand, and attributes. The annotated datasets were pivotal for developing robust product recognition and visual search systems, directly improving the user experience for online shopping platforms. Maintained a consistently high accuracy rate >90% and adhered to strict labeling guidelines to ensure the delivery of reliable, model-ready data.

Executed precise image annotation projects for TELUS Digital to enhance computer vision models for retail and e-commerce applications. Tasks included drawing high-accuracy bounding boxes and polygon segments to identify and isolate thousands of individual products within complex scenes. This involved detailed classification of items based on category, brand, and attributes. The annotated datasets were pivotal for developing robust product recognition and visual search systems, directly improving the user experience for online shopping platforms. Maintained a consistently high accuracy rate >90% and adhered to strict labeling guidelines to ensure the delivery of reliable, model-ready data.

2025

LLM Evaluation & Text Data Generation for Advanced Conversational AI

AppenTextPrompt Response Writing SFT

Served as an AI Training Data Specialist for Outlier.ai, creating and refining high-quality text data to train and evaluate state-of-the-art large language models. The scope involved generating diverse, human-like prompts and responses across a wide range of topics to expand model knowledge and conversational ability. A core responsibility was meticulously evaluating LLM outputs based on criteria such as accuracy, coherence, helpfulness, and potential bias, providing ranked comparisons and detailed feedback. This direct human feedback was critical for fine-tuning model performance and aligning outputs with safety and quality guidelines. The project required a strong command of nuanced English and the ability to research complex topics to ensure factual correctness.

Served as an AI Training Data Specialist for Outlier.ai, creating and refining high-quality text data to train and evaluate state-of-the-art large language models. The scope involved generating diverse, human-like prompts and responses across a wide range of topics to expand model knowledge and conversational ability. A core responsibility was meticulously evaluating LLM outputs based on criteria such as accuracy, coherence, helpfulness, and potential bias, providing ranked comparisons and detailed feedback. This direct human feedback was critical for fine-tuning model performance and aligning outputs with safety and quality guidelines. The project required a strong command of nuanced English and the ability to research complex topics to ensure factual correctness.

2024 - 2025

Education

Work History