Evidence over magic.
Benchmarks, source links, attention maps, compiler output, and traces make decisions reviewable.
About
I’m happiest when the black box becomes understandable.
I’m a computer scientist from the Bay Area beginning graduate study at UC San Diego. My work moves between AI and systems, but the throughline is consistent: make the complicated thing observable, defensible, and useful to someone.
AMBAY AREA → SAN DIEGO
I tend to learn by taking on work before I feel completely ready, then closing the gap by building, tracing, breaking, and rebuilding.
That approach helped me complete my bachelor’s degree in three years while balancing technical projects, hackathons, an internship, coursework, leadership, and work as a Learning Assistant.
My projects look broad on paper—medical imaging, local models, accessibility, security tooling, RAG, Linux, and an operating system—but they keep returning to the same ideas: evidence over magic, local execution where it matters, and interfaces that help people understand what the system is doing.
Arch Linux and Hyprland are my daily environment, and my Windows home server runs an evolving mix of Docker, Jellyfin, Tailscale, remote access, and virtualization. Photography and mechanical watches—especially Spring Drive and Seiko—give me different ways to practice the same attention to systems and details.
How I work
Local when it matters.
Privacy, access, and responsiveness are why several of my AI projects run fully on-device.
Teach the reasoning.
The goal is not only to fix the bug. It is to help someone recognize the next one themselves.
Beyond coursework
Community is part of the work.
I mentored incoming students through MESA, co-led a 30-member recreational math and computer science club, and earlier spent two years leading a Tamil youth organization that raised funds and prepared meals for local homeless shelters. The settings changed; the instinct to make things more approachable did not.