Miriam Moseti

I’m a bilingual STEM professional and AI Training & Prompt Engineer with 6+ years of experience creating high-quality image-based AI training content. I build structured training examples using scientific diagrams, charts, and data visualizations, explaining them with precise technical accuracy in English and Kiswahili.

I’m especially strong in analytical reasoning and QA—interpreting complex data visualizations, detecting logic and annotation errors, and ensuring scientific correctness across bilingual datasets. Through meticulous attention to detail and error detection, I achieved 98% annotation accuracy.

I apply LLM feedback loops for continuous model improvement and deliver consistent results across long-term independent projects. My work spans prompt engineering and response evaluation, structured data creation (JSON, XML, Markdown), and process optimization to keep outputs aligned with human intent.

Alongside professional work, I’m currently pursuing a PhD in Linguistics with a focus on computational language processing. I bring an educator’s mindset—clarifying technical language, supporting multilingual datasets, and using NLP and transcription experience to strengthen training quality and evaluation.