Open to opportunities

Dancan Wanyonyi

@dancanwanyonyi

Message

AI data training specialist focused on high-quality annotation and ML dataset integrity.

Kenya

Message

What I'm looking for

I seek roles in ML data operations or annotation teams where I can improve dataset quality, contribute to LLM evaluation/RLHF, and drive measurable model improvements in a collaborative, fast-paced environment.

I am a detail-oriented AI data training specialist with hands-on experience in data annotation, dataset development, quality assurance, and model evaluation. I have worked across text, image, video, LiDAR and sensor data to support perception and LLM tasks.

My background includes LLM prompt evaluation, RLHF work, and designing annotation workflows that improve model accuracy and reduce labeling errors. I have maintained labeling accuracy above 98% through multi-stage QA processes.

I contributed to autonomous driving projects—annotating 10,000+ frames and improving object-detection accuracy by about 12%—and developed 300+ prompts for safety and reasoning evaluation. I also improved annotation throughput by 25% and reduced common mislabeling by 30% through process feedback.

I hold a Bachelor of Arts in Geography and Environmental Studies and several industry certificates (Google Data Analytics, DeepLearning.AI: AI For Everyone, Udacity Intro to Machine Learning, Coursera: Data Labeling for ML). I am available for remote, hybrid, or contract roles and comfortable with flexible schedules.

Experience

Work history, roles, and key accomplishments

Current

AI Data Training Specialist

Current

CrowdGen

Mar 2020 - Present (5 years 10 months)

Trained LLMs by ranking outputs, evaluating reasoning quality for RLHF pipelines, and creating prompt datasets that improved conversational and domain-specific model performance; maintained labeling accuracy above 98%.

LLM Output Evaluation Prompt Engineering RLHF Pipeline & QA Data Cleaning Labeling Tools

Data Annotation Specialist

Remotask

Jan 2017 - Dec 2019 (2 years 11 months)

Annotated autonomous driving datasets including bounding boxes, segmentation, tracking and LiDAR point clouds; improved annotation throughput by 25% and reduced mislabeling errors by 30%.

& QA Annotation Workflows Labeling Tools