We ship AI that runs on real cameras, not in a demo.
Our software already runs on live government camera grids: thirty feeds on a single laptop, read in real time, with measured accuracy. Everyone here works across the stack, from the model to the operator's screen, and sees their work used in the field.
01
Ship to the field
The work is done when it runs on a real camera at a real site, not when the notebook looks good.
02
Own the outcome
Small team, no hand-offs. You take a problem from the first frame to the operator's decision.
03
Measure everything
We label field data by hand and publish the numbers. If we can't measure it, we don't ship it.
Open roles in India.
No open positions right now.
We're always glad to hear from people who want to build vision AI for the physical world. Tell us what you've built and what you'd like to work on.
Write to us ↗