Open to opportunities

Imperio SHOTTA

@imperioshotta

Message

Senior Yoruba AI training specialist optimizing LLMs with RLHF for culturally accurate Yoruba outputs.

Nigeria

Message

What I'm looking for

I seek roles improving Yoruba/low-resource language performance in LLMs using RLHF, data quality, prompt engineering, and cultural-safety work.

I am a Senior Yoruba AI Training Specialist and RLHF expert with hands-on experience refining Yoruba-language datasets and workflows for global AI labs. I specialize in ensuring cultural resonance, safety, and linguistic accuracy in model responses through meticulous data annotation and evaluation.

My work includes executing RLHF pipelines, red-teaming, adversarial prompting, hallucination detection, and multi-turn prompt engineering to stress-test Yoruba logic, proverbs, and idiomatic expressions. I have consistently maintained 98%+ accuracy across data labeling and QA projects while providing high-fidelity ground truth for model fine-tuning.

I hold a Bachelor of Arts in Yoruba Language and certifications in Advanced RLHF, Python for Data Science, and Prompt Engineering, and I bring a pragmatic, detail-oriented approach to improving low-resource language performance in conversational AI.

Experience

Work history, roles, and key accomplishments

Current

Senior AI Content Specialist

Current

Scale AI

Jan 2024 - Present (2 years 2 months)

Execute RLHF workflows to optimize Yoruba LLM fluency, auditing responses for hallucinations and cultural-safety violations and improving morphological and tone-marking accuracy.

Yoruba Prompt Engineering Data Annotation Low Resource NLP Quality Assurance

Yoruba AI Training Contributor

DataAnnotation.tech

Jun 2023 - Dec 2023 (6 months)

Participated in red-teaming and adversarial prompting for Yoruba-capable models and categorized high-quality Yoruba datasets to improve conversational flow and reduce bias.

Adversarial Prompting Data Annotation Yoruba & QA NLP

Yoruba Language Data Evaluator

Appen

Jan 2022 - May 2023 (1 year 4 months)

Performed entity sentiment analysis and MT evaluation between English and Yoruba, maintaining 98%+ accuracy across data labeling and quality assurance projects.