For employers

Hire this AI Trainer

Sign in or create an account to invite AI Trainers to your job.

LoginCreate account
Invite to Job
Emmanuel Akpan

Emmanuel Akpan

AI Research & Data Annotation Specialist

USA flag
San Francisco, Usa
$5.00/hrIntermediateOtherAppenCVAT

Key Skills

Software

Other
AppenAppen
CVATCVAT
iMeritiMerit
LionbridgeLionbridge
Micro1
OneFormaOneForma
MindriftMindrift
RoboflowRoboflow

Top Subject Matter

General/NLP/AI Domain Expertise
Ecommerce
GUI Data Annotator (Ubuntu/Linux/Python)

Top Data Types

TextText
ImageImage
VideoVideo

Top Task Types

Bounding Box
Segmentation
Classification
Polygon
Polyline
Object Detection
Point Key Point
Question Answering
Text Summarization
Fine Tuning
Evaluation Rating
Computer Programming Coding
Data Collection
Prompt Response Writing SFT
Cuboid
Text Generation
Function Calling

Freelancer Overview

AI Research & Annotation Contributor. Brings 1+ years of professional experience across complex professional workflows, research, and quality-focused execution. Core strengths include Other. Education includes Bachelor of Science, Delta State University (2022). AI-training focus includes data types such as Text and labeling workflows including Evaluation and Rating.

IntermediateFrenchEnglish

Labeling Experience

AI Research & Annotation Contributor

OtherText
This role involved collaborating with generative AI models to evaluate, annotate, and refine text outputs for improved accuracy and user experience. I performed structured research, fact-checking, intent alignment, and guideline compliance checks on AI-generated responses. My work contributed to enhancing the real-world usability and reliability of large language models (LLMs). • Reviewed AI outputs for factual correctness and coherence • Fact-checked claims using credible online sources • Refined prompts and annotated inconsistencies • Identified and corrected hallucinations and reasoning errors.

This role involved collaborating with generative AI models to evaluate, annotate, and refine text outputs for improved accuracy and user experience. I performed structured research, fact-checking, intent alignment, and guideline compliance checks on AI-generated responses. My work contributed to enhancing the real-world usability and reliability of large language models (LLMs). • Reviewed AI outputs for factual correctness and coherence • Fact-checked claims using credible online sources • Refined prompts and annotated inconsistencies • Identified and corrected hallucinations and reasoning errors.

2024 - Present

Product Availability Search for Search Engine Evaluation

ImageEvaluation Rating
I contributed to a search engine evaluation project within the e-commerce industry, focused on assessing the accuracy of product availability search results. The scope involved analyzing thousands of user search queries against actual product inventory to determine whether the search engine returned relevant and available items. The specific tasks included comparing user queries to product titles and attributes, rating result relevance based on predefined scales, and identifying cases where out-of-stock or irrelevant items appeared. In terms of project size, I evaluated approximately 2,500 query-product pairs across multiple categories. Regarding quality measures, I adhered to strict relevance guidelines, participated in regular calibration sessions to maintain rating consistency, and flagged recurring issues to help improve both the annotation process and the overall search experience.

I contributed to a search engine evaluation project within the e-commerce industry, focused on assessing the accuracy of product availability search results. The scope involved analyzing thousands of user search queries against actual product inventory to determine whether the search engine returned relevant and available items. The specific tasks included comparing user queries to product titles and attributes, rating result relevance based on predefined scales, and identifying cases where out-of-stock or irrelevant items appeared. In terms of project size, I evaluated approximately 2,500 query-product pairs across multiple categories. Regarding quality measures, I adhered to strict relevance guidelines, participated in regular calibration sessions to maintain rating consistency, and flagged recurring issues to help improve both the annotation process and the overall search experience.

2025 - 2026

Dynamic Object Tracking for Retail Analytics

VideoBounding Box
I contributed to a video annotation project within the retail analytics industry, focused on training an AI model to analyze customer behavior in physical store environments. The scope involved processing hundreds of hours of surveillance footage, requiring frame-by-frame annotation to track customer movements and interactions. Specific tasks included continuous bounding box tracking of shoppers across sequences, re-identifying subjects upon frame re-entry, and annotating actions like product picking and checkout queue waiting. In terms of project size, I processed approximately 50 hours of footage, resulting in over 180,000 annotated frames covering more than 300 unique customer journeys. Regarding quality measures, I adhered to strict protocols for handling occlusions and lighting variations, participated in regular inter-annotator agreement checks to ensure consistency, and documented edge cases to refine guidelines and enhance the reliability of the training dataset.

I contributed to a video annotation project within the retail analytics industry, focused on training an AI model to analyze customer behavior in physical store environments. The scope involved processing hundreds of hours of surveillance footage, requiring frame-by-frame annotation to track customer movements and interactions. Specific tasks included continuous bounding box tracking of shoppers across sequences, re-identifying subjects upon frame re-entry, and annotating actions like product picking and checkout queue waiting. In terms of project size, I processed approximately 50 hours of footage, resulting in over 180,000 annotated frames covering more than 300 unique customer journeys. Regarding quality measures, I adhered to strict protocols for handling occlusions and lighting variations, participated in regular inter-annotator agreement checks to ensure consistency, and documented edge cases to refine guidelines and enhance the reliability of the training dataset.

2025 - 2025

Semantic Segmentation for Autonomous Vehicle Perception

ImagePolygon
I served as a key contributor to a high-precision data labeling project within the autonomous vehicle industry, focusing on semantic segmentation for road scene perception. The scope of the project involved processing thousands of frames extracted from urban driving footage, requiring meticulous polygon annotation to outline critical visual elements. The specific data labeling tasks performed included tracing the precise boundaries of static objects such as vehicles, pedestrians, traffic signs, and lane markings, as well as dynamic elements like road boundaries and vegetation. Unlike simple bounding boxes, the polygon format demanded a granular level of detail to capture the exact contours of each object, ensuring the model could accurately distinguish shapes and depths. In terms of project size, I successfully annotated over 1,500 complex traffic scenes, contributing to a dataset of roughly 10,000 high-detail instances. Regarding the quality measures adhered to, I maintained strict adherence to a comprehensive guideline that specified vertex count and edge tightness to avoid background bleed, subjected my work to regular auditor cross-checks to ensure geometric precision, and consistently flagged ambiguous frames; such as those with heavy shadows or adverse weather to supervisors to help refine the project ontology and maintain the highest standards of training data integrity.

I served as a key contributor to a high-precision data labeling project within the autonomous vehicle industry, focusing on semantic segmentation for road scene perception. The scope of the project involved processing thousands of frames extracted from urban driving footage, requiring meticulous polygon annotation to outline critical visual elements. The specific data labeling tasks performed included tracing the precise boundaries of static objects such as vehicles, pedestrians, traffic signs, and lane markings, as well as dynamic elements like road boundaries and vegetation. Unlike simple bounding boxes, the polygon format demanded a granular level of detail to capture the exact contours of each object, ensuring the model could accurately distinguish shapes and depths. In terms of project size, I successfully annotated over 1,500 complex traffic scenes, contributing to a dataset of roughly 10,000 high-detail instances. Regarding the quality measures adhered to, I maintained strict adherence to a comprehensive guideline that specified vertex count and edge tightness to avoid background bleed, subjected my work to regular auditor cross-checks to ensure geometric precision, and consistently flagged ambiguous frames; such as those with heavy shadows or adverse weather to supervisors to help refine the project ontology and maintain the highest standards of training data integrity.

2025 - 2025

E-commerce Product Detection & Categorization (Bounding Box Annotation)

ImageBounding Box
From February to May 2025, I contributed to a major data labeling initiative aimed at enhancing a computer vision model for e-commerce applications. The scope of the project involved processing thousands of product images across categories like apparel and electronics, where I was responsible for performing precise bounding box annotations to detect objects in various settings, including studio shots and cluttered backgrounds. My specific tasks included drawing tight bounding boxes around individual items, applying accurate class labels based on a strict taxonomy, and handling complex cases involving occlusion or truncation to ensure the model could learn from real-world conditions. Throughout the project, I adhered to rigorous quality measures by strictly following detailed annotation guidelines, participating in peer review cycles to maintain a high accuracy rate, and proactively flagging edge cases to help refine labeling protocols and improve the overall integrity of the training data.

From February to May 2025, I contributed to a major data labeling initiative aimed at enhancing a computer vision model for e-commerce applications. The scope of the project involved processing thousands of product images across categories like apparel and electronics, where I was responsible for performing precise bounding box annotations to detect objects in various settings, including studio shots and cluttered backgrounds. My specific tasks included drawing tight bounding boxes around individual items, applying accurate class labels based on a strict taxonomy, and handling complex cases involving occlusion or truncation to ensure the model could learn from real-world conditions. Throughout the project, I adhered to rigorous quality measures by strictly following detailed annotation guidelines, participating in peer review cycles to maintain a high accuracy rate, and proactively flagging edge cases to help refine labeling protocols and improve the overall integrity of the training data.

2025 - 2025

Education

D

Delta State University

Bachelor of Science, Computer Science

Bachelor of Science
2018 - 2022

Work History

R

Remote

AI Research & Annotation Contributor (Freelance / Project-Based)

Lagos
2024 - Present
R

Research Institute

Research and Content Analyst

Remote
2023 - 2023