Delight Uzochukwu

Key Skills

Software

Labelbox

Scale AI

Remotasks

Top Subject Matter

AI Evaluation & Model Alignment

Natural Language Understanding (NLU)

Multimodal Data Quality Assurance

Top Data Types

Image

Audio

Text

Top Task Types

Transcription

Classification

Bounding Box

Freelancer Overview

AI Data Quality Assurance Lead (Multimodal MOS Evaluation). Brings 3+ years of professional experience across complex professional workflows, research, and quality-focused execution. Core strengths include Labelbox, Scale AI, and Remotasks. AI-training focus includes data types such as Image, Audio, and Text and labeling workflows including Evaluation, Rating, and Transcription.

Labeling Experience

AI Data Quality Assurance Lead (NLU & Intent Optimization)

LabelboxTextClassification

I performed Natural Language Understanding (NLU) intent classification and optimization for voice assistant models across multilingual markets. I annotated and validated intent data, improving recognition precision and response relevance in AI devices. These activities were crucial for localizing device experiences and adapting to diverse linguistic environments. • Labeled and audited large datasets for intent classification and sentiment analysis tasks. • Conducted linguistic quality assurance across multiple languages and regions. • Collaborated with cross-functional teams to refine AI response logic. • Supported project management and UI translation for localization initiatives.

I performed Natural Language Understanding (NLU) intent classification and optimization for voice assistant models across multilingual markets. I annotated and validated intent data, improving recognition precision and response relevance in AI devices. These activities were crucial for localizing device experiences and adapting to diverse linguistic environments. • Labeled and audited large datasets for intent classification and sentiment analysis tasks. • Conducted linguistic quality assurance across multiple languages and regions. • Collaborated with cross-functional teams to refine AI response logic. • Supported project management and UI translation for localization initiatives.

2024 - Present

AI Data Quality Assurance Lead (Audio QA & Linguistic Curation)

LabelboxAudioTranscription

I supervised and audited the transcription output from external data teams to ensure more than 99% accuracy for speech recognition model training. I implemented and optimized workflows to maximize audio QA throughput and address data inconsistencies effectively. This work helped guarantee high-performance and safe deployment of speech technology at a global scale. • Oversaw linguistic QA processes and established internal audit standards for audio datasets. • Evaluated thousands of hours of speech data for ground truth accuracy and annotation consistency. • Applied advanced workflow optimization to improve throughput metrics by 20%. • Generated evaluation reports and actionable insights for continuous speech model improvement.

I supervised and audited the transcription output from external data teams to ensure more than 99% accuracy for speech recognition model training. I implemented and optimized workflows to maximize audio QA throughput and address data inconsistencies effectively. This work helped guarantee high-performance and safe deployment of speech technology at a global scale. • Oversaw linguistic QA processes and established internal audit standards for audio datasets. • Evaluated thousands of hours of speech data for ground truth accuracy and annotation consistency. • Applied advanced workflow optimization to improve throughput metrics by 20%. • Generated evaluation reports and actionable insights for continuous speech model improvement.

2024 - Present

AI Data Quality Assurance Lead (Multimodal MOS Evaluation)

LabelboxImage

I led Mean Opinion Score (MOS) assessment of multimodal datasets, including images, PDFs, and YouTube media, to calibrate AI model outputs for content recommendation and OCR. My responsibilities included ensuring subjective quality, ground truth verification, and evaluating overall dataset integrity using both manual and software-driven approaches. These evaluations significantly improved accuracy and user experience for millions of device users globally. • Executed image and document-based MOS and subjective quality evaluation projects using multiple toolsets. • Calibrated content recommendation engine performance and OCR systems with high-fidelity ground truthing. • Identified data quality discrepancies and led remediation cycles for faulty or biased data entries. • Delivered feedback to engineering teams for continuous AI optimization.

I led Mean Opinion Score (MOS) assessment of multimodal datasets, including images, PDFs, and YouTube media, to calibrate AI model outputs for content recommendation and OCR. My responsibilities included ensuring subjective quality, ground truth verification, and evaluating overall dataset integrity using both manual and software-driven approaches. These evaluations significantly improved accuracy and user experience for millions of device users globally. • Executed image and document-based MOS and subjective quality evaluation projects using multiple toolsets. • Calibrated content recommendation engine performance and OCR systems with high-fidelity ground truthing. • Identified data quality discrepancies and led remediation cycles for faulty or biased data entries. • Delivered feedback to engineering teams for continuous AI optimization.

2024 - Present

Education

Work History