We build the video calling inside your product.

Usually yes. Most engagements are adding real-time calling to a product that already exists, or replacing a media layer inside one. We work in your repository, with your conventions.

That is one of the most common jobs we take. Typically onto self-hosted LiveKit or mediasoup. The work is the migration path and the load testing — we model your per-minute cost before and after so the decision is made on numbers.

Almost always NAT traversal, a missing or overloaded TURN relay, or a renegotiation bug on reconnect. It is diagnosable — we instrument the ICE candidate flow and the failures usually become obvious within a day or two.

Yes. We build new interfaces to WCAG 2.2 AA and remediate existing ones, testing against the screen readers people actually use — JAWS and NVDA on Windows, VoiceOver on Apple devices — rather than relying on an automated scan. Real-time UIs are the hard case and the one we know best: call controls that announce their state, live regions that report who joined without talking over you, and focus that goes somewhere sensible when a dialog closes.

A call, then a written scope with milestones and a fixed price per milestone. You approve each one before the next begins. Ongoing work is contracted the same way.

You do, on delivery. It lives in your repository throughout, so there is no handover cliff at the end.