For employers

Hire this AI Trainer

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

LoginCreate account
Invite to Job
Joshua Owoh

Joshua Owoh

Data, Image Video labeling and Annotation

JAPAN flag
Tochigi City, Japan
$30.00/hrIntermediateCVATData Annotation TechGoogle Cloud Vertex AI

Key Skills

Software

CVATCVAT
Data Annotation TechData Annotation Tech
Google Cloud Vertex AIGoogle Cloud Vertex AI
LabelboxLabelbox
LabelImgLabelImg
MercorMercor
OneFormaOneForma
TelusTelus
Other
Internal/Proprietary Tooling

Top Subject Matter

No subject matter listed

Top Data Types

AudioAudio
ImageImage
TextText

Top Label Types

Classification
Entity Ner Classification
Evaluation Rating
Question Answering
Segmentation

Freelancer Overview

I am an adaptable and detail-oriented professional with hands-on experience in data entry, transcription, and image/video labeling, particularly within medical and educational domains. My background includes high-volume audio and video transcription, annotation, and subtitling using tools like Aegisub, with a proven track record of maintaining 98% accuracy and strict data privacy. I am proficient in managing and analyzing data, ensuring compliance with industry standards such as HIPAA, and have consistently contributed to efficient data processing and organization. My technical skills span Microsoft Office Suite, Google Workspace, and various electronic record systems, and I am adept at working remotely with optimized setups for seamless communication and collaboration. My experience in medical data management, combined with strong analytical and organizational abilities, enables me to deliver high-quality results in data labeling and AI training data projects.

IntermediateIgboEnglishJapanese

Labeling Experience

Project Kurnakovite – Translation & OCR Quality Evaluation

Internal Proprietary ToolingTextClassificationQuestion Answering
I evaluated ML-based translations and OCR outputs for Meta’s CommsX project by reviewing chat, email, and image text to ensure accuracy, risk-free language, and equivalency to the source material. Tasks included labeling Added Risk (abusive, ideological, legal), assigning XSTS equivalency scores, scoring factualness, voice/tone, grammar, and terminology, and checking OCR text detection and directionality. All work followed strict Kurnakovite guidelines and was completed in SRT HALO/GALA.

I evaluated ML-based translations and OCR outputs for Meta’s CommsX project by reviewing chat, email, and image text to ensure accuracy, risk-free language, and equivalency to the source material. Tasks included labeling Added Risk (abusive, ideological, legal), assigning XSTS equivalency scores, scoring factualness, voice/tone, grammar, and terminology, and checking OCR text detection and directionality. All work followed strict Kurnakovite guidelines and was completed in SRT HALO/GALA.

2025

AI Data Evaluator and Linguistic QA Specialist

Internal Proprietary ToolingTextClassificationEvaluation Rating
I worked on Google’s Data Compute/Gemini evaluation pipeline, reviewing model outputs across text, chat, and multilingual tasks. My role involved scoring responses for factual accuracy, safety, clarity, tone, and instruction-following. I evaluated both single turn and multi turn conversations, checking whether the model stayed on topic and supported the user effectively. I also performed translation quality checks, comparing EN↔JA outputs for equivalency, terminology accuracy, and cultural fit. For factuality tasks, I verified claims using approved sources and applied freshness and stability rules. I handled local-adaptation evaluations, ensuring responses were appropriate for the target region and language norms. In addition, I completed SxS comparisons, choosing the better model output across multiple quality dimensions. Throughout the project, I maintained consistent accuracy and followed Google’s strict rating guidelines to help improve overall Gemini model performance.

I worked on Google’s Data Compute/Gemini evaluation pipeline, reviewing model outputs across text, chat, and multilingual tasks. My role involved scoring responses for factual accuracy, safety, clarity, tone, and instruction-following. I evaluated both single turn and multi turn conversations, checking whether the model stayed on topic and supported the user effectively. I also performed translation quality checks, comparing EN↔JA outputs for equivalency, terminology accuracy, and cultural fit. For factuality tasks, I verified claims using approved sources and applied freshness and stability rules. I handled local-adaptation evaluations, ensuring responses were appropriate for the target region and language norms. In addition, I completed SxS comparisons, choosing the better model output across multiple quality dimensions. Throughout the project, I maintained consistent accuracy and followed Google’s strict rating guidelines to help improve overall Gemini model performance.

2023

Avocado Web Parser – Data Labeling & Quality Annotation

Internal Proprietary ToolingTextClassificationEvaluation Rating
I reviewed large scale webpage text and HTML to classify content quality and identify issues such as generated content, spam, clone sites, and structured pages. I also compared crawled text against the original webpage to flag missing, broken, or unnecessary content. All tasks were completed in SRT HALO/GALA using strict guidelines to ensure clean, accurate pretraining data.

I reviewed large scale webpage text and HTML to classify content quality and identify issues such as generated content, spam, clone sites, and structured pages. I also compared crawled text against the original webpage to flag missing, broken, or unnecessary content. All tasks were completed in SRT HALO/GALA using strict guidelines to ensure clean, accurate pretraining data.

2023

LLM Text-2-image

Internal Proprietary ToolingTextQuestion AnsweringText Generation
I evaluated AI-generated images by reviewing how well each output matched the written prompt and whether it met basic visual quality standards. For every comparison, I judged which image aligned better with the prompt’s intent, clarity, subject accuracy, style cues, and overall coherence. I also assessed visual elements such as detail preservation, artifact presence, composition, and whether the image felt complete or distorted. Working with a structured rubric, I selected A, B, or Tie depending on how well each image reflected the prompt and maintained visual quality. The role required steady judgment, strong attention to detail, and the ability to spot major mismatches such as off-prompt objects, incorrect attributes, or structural errors while also noticing subtle differences where both images were close. These evaluations directly supported the development of text to-image models, helping improve prompt alignment and reduce obvious generation mistakes. The project strengthened my

I evaluated AI-generated images by reviewing how well each output matched the written prompt and whether it met basic visual quality standards. For every comparison, I judged which image aligned better with the prompt’s intent, clarity, subject accuracy, style cues, and overall coherence. I also assessed visual elements such as detail preservation, artifact presence, composition, and whether the image felt complete or distorted. Working with a structured rubric, I selected A, B, or Tie depending on how well each image reflected the prompt and maintained visual quality. The role required steady judgment, strong attention to detail, and the ability to spot major mismatches such as off-prompt objects, incorrect attributes, or structural errors while also noticing subtle differences where both images were close. These evaluations directly supported the development of text to-image models, helping improve prompt alignment and reduce obvious generation mistakes. The project strengthened my

2023

Video Quality Compare

Don T DiscloseVideoSegmentationEmotion Recognition
I worked as a video quality evaluator for an AI video-generation and reconstruction project. My job was to compare two reconstructed video clips against the original and decide which one delivered the closest match. Each decision required close attention to motion consistency, frame-to-frame stability, visual clarity, artifact detection, and how well each reconstruction preserved the source video’s details and timing. Using a defined rubric, I selected A, B, or Tie based on overall similarity and technical quality. The role demanded precise judgment, consistency across large volumes of comparisons, and the ability to spot subtle differences in color shifts, compression artifacts, motion blur, and temporal distortions. These evaluations were used to guide model improvements and refine how reliably the system reproduced high-quality video. This project strengthened my ability to assess complex visual sequences, make objective decisions in borderline cases, and maintain accuracy while r

I worked as a video quality evaluator for an AI video-generation and reconstruction project. My job was to compare two reconstructed video clips against the original and decide which one delivered the closest match. Each decision required close attention to motion consistency, frame-to-frame stability, visual clarity, artifact detection, and how well each reconstruction preserved the source video’s details and timing. Using a defined rubric, I selected A, B, or Tie based on overall similarity and technical quality. The role demanded precise judgment, consistency across large volumes of comparisons, and the ability to spot subtle differences in color shifts, compression artifacts, motion blur, and temporal distortions. These evaluations were used to guide model improvements and refine how reliably the system reproduced high-quality video. This project strengthened my ability to assess complex visual sequences, make objective decisions in borderline cases, and maintain accuracy while r

2023

Education

O

Our Lady of Fatima University

Bachelor of Science in Nursing (BSN), Nursing

Bachelor of Science in Nursing (BSN)
2012 - 2017
O

Our Lady of Fatima University

Bachelor of Science, Nursing

Bachelor of Science
2014 - 2016

Work History

S

SpeakBuddy

Online English Conversation Coach — Remote

Tochigi
2025 - Present
F

Freelancer

Search Quality Rater & Data Annotator

Tochigi
2024 - Present