At Fogsphere, I design CI/CD pipelines that take machine learning models from training and evaluation through versioning in reproducible, containerized environments. I’ve also built reusable CLI tools for dataset processing and experiment execution.
I developed the Finance Portal, a local-first personal finance app published on the Microsoft Store, with an LLM assistant that answers from tool outputs over local financial data. I secured its lock screen with authenticated encryption and packaged the app for distribution.
For my M.Sc. thesis, I built Computer Vision Studio, a desktop platform for auto-labeling, training, and evaluating computer vision models. Its auto-labeling processed images in 0.14 seconds, compared with 0.6 seconds in Roboflow and 9 seconds in CVAT.
I also developed Examus, a full-stack exam-generation platform now in production at Universidad Mayor de San Andrés. My work spans software and MLOps, computer vision, LLM applications, and robotics; I have an M.Sc. in Mechatronic Systems and a Springer-published research paper.
