Barnabas Chukwuemeka

Key Skills

Software

Appen

Mindrift

OneForma

Toloka

Telus

Other

Top Subject Matter

No subject matter listed

Top Data Types

Image

Text

Video

Top Task Types

Classification

Entity Ner Classification

Evaluation Rating

Mapping

Prompt Response Writing SFT

Freelancer Overview

I’m an experienced AI Data Specialist with over 2 years of hands-on work in data labeling, content evaluation, and AI training. I’ve contributed to projects involving AI response rating, search quality evaluation, and social media content moderation (Appen), consistently applying complex guidelines to ensure accuracy, consistency, and high-quality data outputs. My work has supported the development of large language models by evaluating factuality, relevance, tone, and harmfulness of AI-generated content. In addition to text-based tasks, I’ve worked with video annotation, assessing subject movement, spatial relationships, and temporal sequences. I’m proficient in tools like Raterhub, UHRS, SRT, and proprietary annotation platforms used by Mindrift and Telus International. My strengths include a deep understanding of user intent, critical thinking, and the ability to handle nuanced annotation scenarios. With a strong eye for detail and a passion for AI-human interaction, I bring both quality and adaptability to every task.

Labeling Experience

Video Search Evaluation – Apple TV

OtherTextClassificationEvaluation Rating

In this evaluation project for Apple TV’s video search system, I rated search result relevance based on complex multi-aspect queries (genre, actor, time period, similarity, etc.). Tasks required interpreting primary and secondary intent, identifying appropriate content (e.g., TV shows, movies), and providing concise reasoning for each rating. I assessed whether results met query intent across browsing, navigational, and similarity categories, using resources like IMDb, Wikipedia, and YouTube. My work helped improve AI-driven search algorithms by refining media relevance and intent handling in video search systems.

In this evaluation project for Apple TV’s video search system, I rated search result relevance based on complex multi-aspect queries (genre, actor, time period, similarity, etc.). Tasks required interpreting primary and secondary intent, identifying appropriate content (e.g., TV shows, movies), and providing concise reasoning for each rating. I assessed whether results met query intent across browsing, navigational, and similarity categories, using resources like IMDb, Wikipedia, and YouTube. My work helped improve AI-driven search algorithms by refining media relevance and intent handling in video search systems.

2024 - 2024

Map Search Evaluation – Local Business & POI Accuracy

TelusGeospatial Tiled ImageryClassificationEvaluation Rating

As part of a rigorous qualification-based evaluation project, I reviewed over 50 real-world map search queries to assess the accuracy and relevance of local business results on a map interface. Tasks involved verifying business presence, name, address, pin accuracy, and correct categorization. I used tools like Google Maps, Street View, official websites, and business directories to ensure that each result reflected real-world conditions. My annotations directly contributed to improving local search results and the navigation experience for users.

As part of a rigorous qualification-based evaluation project, I reviewed over 50 real-world map search queries to assess the accuracy and relevance of local business results on a map interface. Tasks involved verifying business presence, name, address, pin accuracy, and correct categorization. I used tools like Google Maps, Street View, official websites, and business directories to ensure that each result reflected real-world conditions. My annotations directly contributed to improving local search results and the navigation experience for users.

2024 - 2024

PII Tagging, Prompt Design & Harm Classification (Qualification Task)

OtherTextEntity Ner ClassificationText Generation

As part of a qualification project for sensitive AI training, I created and evaluated structured prompts containing implicit and explicit PII across various domains (Insurance, Finance, Safety). Tasks involved inserting domain-specific and general PII (e.g., addresses, passwords), tagging harm categories (e.g., Jailbreak, Violence, Insult), and crafting prompts that mimic malicious and non-malicious intent. I demonstrated the ability to handle red teaming scenarios, implicit harmful use cases, and ethical data creation under strict labeling instructions. All prompts followed guidelines for structure, tone, content length, and intent. I passed this qualification with strong accuracy across all required labeling dimensions.

As part of a qualification project for sensitive AI training, I created and evaluated structured prompts containing implicit and explicit PII across various domains (Insurance, Finance, Safety). Tasks involved inserting domain-specific and general PII (e.g., addresses, passwords), tagging harm categories (e.g., Jailbreak, Violence, Insult), and crafting prompts that mimic malicious and non-malicious intent. I demonstrated the ability to handle red teaming scenarios, implicit harmful use cases, and ethical data creation under strict labeling instructions. All prompts followed guidelines for structure, tone, content length, and intent. I passed this qualification with strong accuracy across all required labeling dimensions.

2024 - 2024

Video Annotation

OtherVideoRelationshipClassification

I annotated short video clips by identifying subjects, describing motion sequences, and marking spatial relationships between people, objects, and the camera. Tasks included labeling body part movements, action sequences, character entry/exit points, and changes in facial expressions. I followed strict project guidelines to maintain objectivity and time-aligned accuracy. The annotation work supported training models in human activity recognition, scene understanding, and motion prediction.

I annotated short video clips by identifying subjects, describing motion sequences, and marking spatial relationships between people, objects, and the camera. Tasks included labeling body part movements, action sequences, character entry/exit points, and changes in facial expressions. I followed strict project guidelines to maintain objectivity and time-aligned accuracy. The annotation work supported training models in human activity recognition, scene understanding, and motion prediction.

2024 - 2024

Search Quality Rating

TelusTextClassificationEvaluation Rating

I evaluated web search results for relevance, intent alignment, and content quality across various query types (informational, transactional, YMYL, etc.). Using platforms like Raterhub, I classified queries, rated SERP relevance, and applied E-E-A-T (Experience, Expertise, Authoritativeness, Trustworthiness) principles. I worked on thousands of query-page pairs, achieving consistent quality scores. Tasks required critical thinking, awareness of time sensitivity, and the ability to judge page trustworthiness based on user intent.

I evaluated web search results for relevance, intent alignment, and content quality across various query types (informational, transactional, YMYL, etc.). Using platforms like Raterhub, I classified queries, rated SERP relevance, and applied E-E-A-T (Experience, Expertise, Authoritativeness, Trustworthiness) principles. I worked on thousands of query-page pairs, achieving consistent quality scores. Tasks required critical thinking, awareness of time sensitivity, and the ability to judge page trustworthiness based on user intent.

2024 - 2024

Show more

Education

Work History