Kelvin Njenga

Key Skills

Software

Labelbox

Prodigy

CVAT

Doccano

Top Subject Matter

Computer Vision

Content Moderation

Autonomous Vehicles

Top Data Types

Image

Text

Document

Top Task Types

Object Detection

Classification

Segmentation

Freelancer Overview

Data Annotation Specialist. Core strengths include Labelbox, Prodigy, and CVAT. Education includes Bachelor of Science, University of Nairobi (2022). AI-training focus includes data types such as Image and Text and labeling workflows including Object Detection, Classification, and Segmentation.

Labeling Experience

AI Training Dataset Curator – Autonomous Vehicle Perception

CVATImageSegmentation

I led segmentation labeling for over 2,000 street scene images to support autonomous vehicle perception. This involved detailed segmentation of occluded objects and images under adverse conditions. I resolved annotation conflicts through consensus and robust escalation protocols. • Utilized CVAT for image annotation tasks. • Improved accuracy in street scene segmentation. • Contributed to high-quality training data for AV perception models. • Addressed challenging scenarios in computer vision datasets.

I led segmentation labeling for over 2,000 street scene images to support autonomous vehicle perception. This involved detailed segmentation of occluded objects and images under adverse conditions. I resolved annotation conflicts through consensus and robust escalation protocols. • Utilized CVAT for image annotation tasks. • Improved accuracy in street scene segmentation. • Contributed to high-quality training data for AV perception models. • Addressed challenging scenarios in computer vision datasets.

2023 - Present

Multilingual Text Classification Specialist

DoccanoTextClassification

I managed multilingual text classification for over 10,000 customer support tickets focusing on intent identification. My work ensured consistent application of taxonomy across different languages. I recognized and incorporated cultural nuances to enhance classification accuracy. • Used Doccano and Prodigy as primary annotation software. • Contributed to LLM training data for NLP tasks. • Enhanced model performance on multilingual datasets. • Implemented robust review and validation procedures.

I managed multilingual text classification for over 10,000 customer support tickets focusing on intent identification. My work ensured consistent application of taxonomy across different languages. I recognized and incorporated cultural nuances to enhance classification accuracy. • Used Doccano and Prodigy as primary annotation software. • Contributed to LLM training data for NLP tasks. • Enhanced model performance on multilingual datasets. • Implemented robust review and validation procedures.

2021 - Present

Data Annotation Specialist

LabelboxImageObject Detection

As a Data Annotation Specialist, I annotated over 15,000 image assets for object detection models across platforms including Remotasks, OneForma, and Mercor. I developed comprehensive annotation guidelines for edge cases, resulting in fewer reviewer escalations. My collaboration with ML engineers helped refine labeling taxonomies based on feedback. • Maintained a 98.5% inter-annotator agreement. • Audited peer submissions and trained new team members. • Used Labelbox, Amazon SageMaker Ground Truth, and CVAT. • Processed datasets for model improvement in image-based AI.

As a Data Annotation Specialist, I annotated over 15,000 image assets for object detection models across platforms including Remotasks, OneForma, and Mercor. I developed comprehensive annotation guidelines for edge cases, resulting in fewer reviewer escalations. My collaboration with ML engineers helped refine labeling taxonomies based on feedback. • Maintained a 98.5% inter-annotator agreement. • Audited peer submissions and trained new team members. • Used Labelbox, Amazon SageMaker Ground Truth, and CVAT. • Processed datasets for model improvement in image-based AI.

2021 - Present

Content Reviewer & Data Validator

ProdigyTextClassification

As a Content Reviewer & Data Validator, I evaluated user-generated text content for policy compliance. I flagged nuanced violations that required human judgement and documented labeling inconsistencies. My work contributed to refining annotation guidelines and training processes. • Achieved 99.2% accuracy compliance. • Used Prodigy and Doccano for data annotation. • Trained five new team members on standards. • Helped improve quality in text annotation for moderation tasks.

As a Content Reviewer & Data Validator, I evaluated user-generated text content for policy compliance. I flagged nuanced violations that required human judgement and documented labeling inconsistencies. My work contributed to refining annotation guidelines and training processes. • Achieved 99.2% accuracy compliance. • Used Prodigy and Doccano for data annotation. • Trained five new team members on standards. • Helped improve quality in text annotation for moderation tasks.

2021 - 2022

Education

Work History