Adebusola Odusina

Key Skills

Software

Appen

Toloka

Other

Mindrift

Top Subject Matter

No subject matter listed

Top Data Types

Audio

Image

Text

Top Label Types

Audio Recording

Classification

Text Generation

Translation Localization

Freelancer Overview

I have over one year of hands-on experience in data labeling and AI training, where I’ve worked on diverse projects involving text classification, prompt evaluation, and image annotation. My tasks have included organizing and labeling large datasets with accuracy, following detailed guidelines, and using tools to train machine learning models effectively. I’ve contributed to improving AI outputs by ensuring high-quality, contextually relevant data. What sets me apart is my keen attention to detail, consistency, and ability to quickly grasp complex instructions. I’ve worked across industries such as education, general AI systems, and visual data projects, and I’m comfortable working both independently and collaboratively. My background in administrative roles also equips me with strong organizational and communication skills, which support my efficiency in meeting deadlines and delivering reliable training data

Labeling Experience

Prompt classification

MindriftTextEvaluation RatingPrompt Response Writing SFT

I participated in a prompt classification project designed to enhance the performance of AI conversational models. The scope of the project involved evaluating and labeling user prompts based on categories such as intent, clarity, tone, complexity, and task type. This helped improve the AI's ability to understand and respond appropriately to various types of user input. The project size covered thousands of prompts across diverse topics including education, daily life, customer service, and technical queries. Each prompt required careful reading, accurate labeling, and sometimes multiple tags depending on the content and purpose. To maintain quality, I followed a comprehensive set of annotation guidelines and underwent frequent calibration sessions to ensure consistency with team standards. My work was subject to regular reviews, and I consistently achieved high accuracy and reliability scores. I also adapted quickly to guideline updates and contributed feedback to improve.

I participated in a prompt classification project designed to enhance the performance of AI conversational models. The scope of the project involved evaluating and labeling user prompts based on categories such as intent, clarity, tone, complexity, and task type. This helped improve the AI's ability to understand and respond appropriately to various types of user input. The project size covered thousands of prompts across diverse topics including education, daily life, customer service, and technical queries. Each prompt required careful reading, accurate labeling, and sometimes multiple tags depending on the content and purpose. To maintain quality, I followed a comprehensive set of annotation guidelines and underwent frequent calibration sessions to ensure consistency with team standards. My work was subject to regular reviews, and I consistently achieved high accuracy and reliability scores. I also adapted quickly to guideline updates and contributed feedback to improve.

2024

Audio - Text classification

OtherAudioText GenerationTranslation Localization

I worked on an audio-to-text transcription project aimed at improving the accuracy of AI speech recognition systems. The project involved transcribing a large volume of diverse audio files—ranging from conversational dialogue to formal speech—while adhering to strict formatting and linguistic accuracy guidelines. The scope included accurately capturing speaker intent, punctuation, speaker turns, and background context to produce clean, structured transcripts. The project size included thousands of audio clips of varying lengths (from 30 seconds to several minutes), often requiring attention to different accents, dialects, and noise levels. To ensure quality, I followed detailed transcription protocols including spelling consistency, non-verbal cue tagging (e.g., pauses, laughter), and timestamping where required. My work underwent regular peer reviews and quality audits, and I consistently met or exceeded the expected accuracy benchmarks (typically above 95%).

I worked on an audio-to-text transcription project aimed at improving the accuracy of AI speech recognition systems. The project involved transcribing a large volume of diverse audio files—ranging from conversational dialogue to formal speech—while adhering to strict formatting and linguistic accuracy guidelines. The scope included accurately capturing speaker intent, punctuation, speaker turns, and background context to produce clean, structured transcripts. The project size included thousands of audio clips of varying lengths (from 30 seconds to several minutes), often requiring attention to different accents, dialects, and noise levels. To ensure quality, I followed detailed transcription protocols including spelling consistency, non-verbal cue tagging (e.g., pauses, laughter), and timestamping where required. My work underwent regular peer reviews and quality audits, and I consistently met or exceeded the expected accuracy benchmarks (typically above 95%).

2024

Education

Work History