Missing Person detection (AI)

When someone goes missing, the search depends on people manually comparing a photograph against faces in footage or in crowds. It is slow, it does not scale past a few hundred images, and attention degrades exactly when it matters most.

Built the Next.js and TypeScript interface over the recognition service — upload, case management and the results view.

The interface treats a match as a candidate, never as an answer. Recognition returns a confidence score, and presenting a probable match as a confirmed identity would be actively dangerous in this domain, so results are ranked and shown with their scores for a human to judge.

Uploads are handled asynchronously with progress surfaced in the interface, since inference over a batch of images takes long enough that a blocking request would time out.

Shipped as FINDER, a face-recognition tool for missing-person cases.