Michael Akinbodewa

Key Skills

Software

Scale AI

Labelbox

CVAT

Label Studio

Top Subject Matter

LLM Alignment

Rlhf Domain Expertise

Chain-of-Thought Reasoning

Top Data Types

Text

Image

Video

Top Task Types

RLHF

Segmentation

Freelancer Overview

Senior AI Content Specialist (LLM & RLHF). Core strengths include Scale AI, Labelbox, and CVAT. Education includes Bachelor of Science, University of Port Harcourt (2021). AI-training focus includes data types such as Text, Image, and Video and labeling workflows including RLHF and Segmentation.

Labeling Experience

Senior AI Content Specialist (LLM & RLHF)

Scale AITextRLHF

As Senior AI Content Specialist at Global AI Labs, I led annotation and QA for large-scale LLM alignment projects, specializing in RLHF rankings and Chain-of-Thought reasoning. I designed annotation guidelines that significantly improved inter-annotator agreement and successfully reduced model hallucination rates. My work included dataset versioning and mentoring junior annotators for sustained quality across model iterations. • Produced and validated over 100,000 RLHF rankings (Helpfulness, Honesty, Harmlessness). • Implemented semi-automated evaluation pipelines with Python and prompt engineering. • Led red-teaming for a 20,000-prompt safety dataset improving model safety benchmarks. • Standardized onboarding for annotators, cutting ramp-up time in half.

As Senior AI Content Specialist at Global AI Labs, I led annotation and QA for large-scale LLM alignment projects, specializing in RLHF rankings and Chain-of-Thought reasoning. I designed annotation guidelines that significantly improved inter-annotator agreement and successfully reduced model hallucination rates. My work included dataset versioning and mentoring junior annotators for sustained quality across model iterations. • Produced and validated over 100,000 RLHF rankings (Helpfulness, Honesty, Harmlessness). • Implemented semi-automated evaluation pipelines with Python and prompt engineering. • Led red-teaming for a 20,000-prompt safety dataset improving model safety benchmarks. • Standardized onboarding for annotators, cutting ramp-up time in half.

2023 - Present

Data Annotation Specialist (Multimodal AI)

LabelboxImageSegmentation

At Innovative Data Solutions, I executed high-volume labeling projects across image, text, and video domains, with a primary focus on segmentation and logic prompt annotation. I created comprehensive annotation specs and training documents to support multimodal LLMs, resulting in improved model validation accuracy. Additionally, I led batch QA processes and spearheaded inter-annotator agreement studies to enhance labeling consistency. • Labeled 120,000+ images with segmentation/masks and 25,000+ logic prompts. • Developed annotation spec docs and visual-QA training materials. • Converted legacy annotation formats and automated dataset standardization with Python. • Presented agreement study findings to improve guideline clarity for edge cases.

At Innovative Data Solutions, I executed high-volume labeling projects across image, text, and video domains, with a primary focus on segmentation and logic prompt annotation. I created comprehensive annotation specs and training documents to support multimodal LLMs, resulting in improved model validation accuracy. Additionally, I led batch QA processes and spearheaded inter-annotator agreement studies to enhance labeling consistency. • Labeled 120,000+ images with segmentation/masks and 25,000+ logic prompts. • Developed annotation spec docs and visual-QA training materials. • Converted legacy annotation formats and automated dataset standardization with Python. • Presented agreement study findings to improve guideline clarity for edge cases.

2022 - 2023

Annotation Lead (Autonomous Perception Segmentation)

CVATVideoSegmentation

I served as annotation lead for autonomous perception projects, focusing on segmentation of video frames to aid model development. My work improved inter-annotator agreement and contributed to significant gains in model mean average precision (mAP). I developed refined guidelines for edge cases and trained team members on annotation best practices. • Annotated 40,000 video frames for object segmentation. • Achieved inter-annotator agreement of 0.88 across the team. • Enhanced model mAP by 15% via improved labeling quality. • Implemented documentation and conducted team training sessions.

I served as annotation lead for autonomous perception projects, focusing on segmentation of video frames to aid model development. My work improved inter-annotator agreement and contributed to significant gains in model mean average precision (mAP). I developed refined guidelines for edge cases and trained team members on annotation best practices. • Annotated 40,000 video frames for object segmentation. • Achieved inter-annotator agreement of 0.88 across the team. • Enhanced model mAP by 15% via improved labeling quality. • Implemented documentation and conducted team training sessions.

Not specified

Education

Work History