I build AI systems that ship: data pipelines, retrieval layers, model APIs, and the product surfaces that make them usable. I care about reliability, clarity, and outcomes that hold up in production.
I enjoy ambiguous problems where the hardest part is shaping the system, not just writing code. I build the smallest reliable version, measure it, and then iterate quickly.
When I am not working, I tinker with Linux and KDE, which is how my plasma applet fork started.
AI infra, full-stack AI products, and MLOps systems where reliability and UX matter as much as the model itself.
Explore my projects