About

I'm a computer science undergraduate at Oregon State University. I build AI systems, and I'm drawn to LLM reliability: what happens to large language models when the hardware under them fails, and how inference can defend itself.

This summer I'm building production AI at Mibanco (part of Credicorp, Peru's largest financial group): a speech-to-text and LLM-interpretation pipeline on Databricks and Azure AI. Before that, at OSU's TRUE (Trustworthy and Responsible) AI Lab, I ran fault-injection experiments on 1B–8B models and did the initial investigation and early results on decoding-time defenses against bit-flip attacks.

I grew up in Lima, Peru, and study CS at Oregon State. I build my own tools, too; the latest is an internship agent I use every morning. And I use AI tooling daily, for development and for staying organized.