Aarav Shah
LungLens is my independent student project. I wanted a calmer way to look again, not a machine that claims to know more than it does.
The numbers that stayed with me are on the overview.
I set the rule that LungLens measures finger shape, not disease. If the photo is poor, I reject it. I keep landmark finding separate from geometry. If I am not confident, I say I cannot tell. Until models are validated, I do not let a stub look clinical. I also shaped the patient, clinician, and admin workflows, the tests, and this site.
Dhaval Shah mentored me on process, review, and staying honest about claims. The architecture and final decisions stayed mine.
I used TypeScript, Next.js, React, Tailwind, Python, FastAPI, tests, Postgres-backed storage, Docker, Git, and Cursor. Assistants helped with boilerplate, debugging, and drafts. I still decided requirements, architecture, and what I call validated versus experimental.
Once landmarks exist, the geometry is the authority. A stub may not show a clinical label. Low confidence must abstain. Patient, clinician, and admin views stay separate. This site only claims what my tests support.