On assignment at Meta through Tundra Technical Solutions.

Evaluated LLM responses for visual content description, ensuring accuracy, truthfulness, and alignment with source imagery. Created 50 meticulously observed visual descriptions, analyzing text placement, typography, layout, and style attributes. Contributed these descriptions to the model’s training dataset, directly improving its visual understanding accuracy from 65% to 87%. Served as a visual annotation specialist, delivering high-quality evaluation inputs that enhanced overall model performance in vision–language tasks.

Worked in creative writing for AI models, assessing and refining model-generated responses in a bilingual environment. This role involved evaluating content for clarity, tone, cultural relevance, and linguistic accuracy, then providing improved, high-quality alternatives to strengthen model performance. By generating refined bilingual outputs and identifying gaps in language understanding, I contributed to more natural, contextually accurate, and user-aligned AI interactions.

As an AI Quality and Localization within an AI/Assistant linguistics team, I contributed to enhancing the performance and cultural relevance of Gemini. My responsibilities included evaluating and refining AI-driven interactions, such as setting reminders, managing tasks, launching various applications, and executing commands with a deep understanding of local language nuances and cultural context. By ensuring that Gemini’s responses were accurate, natural, and contextually appropriate, I helped improve its ability to provide seamless, personalized user experiences across diverse linguistic and cultural backgrounds. This role required meticulous attention to detail, strong linguistic skills, and the ability to focus on strengthening the relationship between AI-driven tools and the end user.

At Gaio, contributed as a UX designer to the redesign of Trash to Cash’s website, a waste management company. Conducted in-depth user research interviews and collaborated closely with clients and cross-functional teams to uncover user needs and business goals. Developed multiple user-centered design concepts focused on creating a fresh, visually engaging website experience that also increased traffic and awareness of Trash to Cash’s mission.