Victor Okonkwo

Key Skills

Software

Data Annotation Tech

Humanatic

Labelbox

Lionbridge

Mercor

Telus

Scale AI

Remotasks

Top Subject Matter

No subject matter listed

Top Data Types

Audio

Document

Image

Top Task Types

Action Recognition

Classification

Data Collection

Text Generation

Text Summarization

Freelancer Overview

Experienced AI data labeling specialist with a strong background in preparing and improving training data for machine learning and artificial intelligence projects. Skilled in working with images, text, audio, and video data using tools such as Labelbox, CVAT, SuperAnnotate, Scale AI, and AWS SageMaker Ground Truth. Over the years, I have contributed to projects involving computer vision, text evaluation, and geospatial analysis by ensuring every dataset meets the highest standards of accuracy and consistency. Known for being detail oriented, reliable, and quick to adapt to new tools and instructions. I have worked closely with AI teams and researchers to design better labeling guidelines, review data for quality, and help build models that perform more effectively. My focus is always on producing clean, well-structured data that strengthens the foundation of AI systems and supports real-world applications.

Labeling Experience

LLM Safety and Quality Evaluation — Outlier Platform

Scale AITextQuestion AnsweringText Summarization

Reviewed and improved model responses for general knowledge, reasoning, and writing quality. Wrote prompts, rated answers, and suggested edits to increase clarity, factual accuracy, and safety. Performed red teaming to probe for harmful or biased outputs, then documented failure cases for guideline updates. Maintained strong reviewer scores while adapting to frequent instruction changes and tight turnarounds.

Reviewed and improved model responses for general knowledge, reasoning, and writing quality. Wrote prompts, rated answers, and suggested edits to increase clarity, factual accuracy, and safety. Performed red teaming to probe for harmful or biased outputs, then documented failure cases for guideline updates. Maintained strong reviewer scores while adapting to frequent instruction changes and tight turnarounds.

2023 - 2025

Ecommerce Catalog Enrichment and Product Image Tagging

LabelboxImageBounding BoxEntity Ner Classification

Tagged product images and enriched item descriptions to improve search relevance, recommendations, and browse quality for a large online retailer. Labeled brand, color, material, and style attributes from images and titles. Drew tight boxes on logos and distinguishing features when needed for visual search. Normalized attributes to a strict taxonomy and flagged duplicates and near duplicates. Reviewed search results and rated relevance to user intent which raised top result precision and reduced zero result queries. Processed about one thousand items per week with steady reviewer approval above ninety eight percent. Built a short playbook for edge cases like multi pack listings and seasonal bundles that cut rework by a third and sped up onboarding for new labelers.

Tagged product images and enriched item descriptions to improve search relevance, recommendations, and browse quality for a large online retailer. Labeled brand, color, material, and style attributes from images and titles. Drew tight boxes on logos and distinguishing features when needed for visual search. Normalized attributes to a strict taxonomy and flagged duplicates and near duplicates. Reviewed search results and rated relevance to user intent which raised top result precision and reduced zero result queries. Processed about one thousand items per week with steady reviewer approval above ninety eight percent. Built a short playbook for edge cases like multi pack listings and seasonal bundles that cut rework by a third and sped up onboarding for new labelers.

2022 - 2024

Autonomous Driving Perception — Vision and LiDAR Annotation (Remotasks / Scale)

Scale AI3D SensorBounding BoxPolygon

Labeled street scenes for vehicle, pedestrian, cyclist, and road-feature detection. Drew lane lines and road boundaries, created polygons for traffic signs and signals, and tracked moving objects across video. Contributed to LiDAR tasks by confirming object extents and consistency with camera frames. Followed gold-task checks and spot reviews to keep accuracy high and rework low.

Labeled street scenes for vehicle, pedestrian, cyclist, and road-feature detection. Drew lane lines and road boundaries, created polygons for traffic signs and signals, and tracked moving objects across video. Contributed to LiDAR tasks by confirming object extents and consistency with camera frames. Followed gold-task checks and spot reviews to keep accuracy high and rework low.

2019 - 2021

Education

Work History