Building intelligent,
accessible technology.

We research the real problem, then build for it.
AI · computer vision · embedded systems.

capabilities

RESEARCH

We start with people.

Before we write any code, we go and talk to the people who would actually use it. The problem usually looks different up close.

FULL STACK

We own the whole thing.

From training a model to the hardware it runs on to the app that talks to it. Keeping it in one team means nothing falls through the gaps.

PROTOTYPING

We build to find out.

A prototype that stays in the lab isn't a prototype. We test in the field, with real users, and let that change the direction.

ACCESSIBILITY

We design for the edge case.

We pick the user with the least margin for error and design from there. It makes the work harder and the result better.

flagship project

ACTIVE · RUNSIGHT · 2025–PRESENT

AI glasses for
visually impaired runners.

“RunSight didn’t start with a piece of technology. It started with a friend who lost their sight and still wanted to run.”

AIComputer VisionEmbedded SystemsAccessibility

View full project →Visit runsight.labmino.com ↗

process

We start with people, not tech. RunSight's biggest turn didn't come off a whiteboard. It came from a conversation, when we realized the problem wasn't getting around in general. It was running at full speed without being able to see the track.

We build to find out, not to be right. When our first model couldn't keep up with how much a track changes, we didn't patch it. We pulled the pipeline apart and rebuilt it as a few models working together. Fixing the approach mattered more than defending it.

Safety is the constraint, not a feature. RunSight is talking to a visually impaired runner at full speed, and a wrong call can't be taken back mid-stride. So the careful parts, the thresholds and the smoothing and the habit of staying quiet when unsure, aren't the finish. They're where we start.

recognition