Skip to content

We have no open roles right now.

We hire when we meet someone who has built the hard part before. If that is you, the address is at the foot of this page and a paragraph beats a covering letter.

Write to the hiring teamSee what we build
  • On-premises runtime
  • Speech and language models
  • Regulated deployments
Five white slabs stand in a row on a bench, each with a coloured edge; the middle one is open on a hinge with its lattice showing and a tool across it, a lens leans over it, and a stack of blank slabs waits at the end.
  1. Five layersTelephony, channel, orchestration, model, infrastructure
  2. The hard partEvery layer has one, and this is where it is
  3. Still openThe work nobody has done yet, which is the job
The runtime

Five layers, and the open one is the job.

What you would work on01 / 04

Five layers, and the hard part is in all of them.

Latency, models, retrieval and a deployment on hardware we do not own. The last one shapes the rest.

Inside someone else’s perimeter
A live call
01

Telephony and media edge

SIP, WebRTC and codecs, on a network you do not control.

Runs on the customer’s own trunks

Write about this layer
02

Turn taking and latency

Barge-in, endpointing and a budget a caller can hear.

Measured on every build

Write about this layer
03

Speech and language

Accents, code-mixed audio and voices, tuned per locale.

Evaluated per locale before release

Write about this layer
04

Retrieval and grounding

A cited answer or an honest no, inside a permission filter.

An answer cites, or it declines

Write about this layer
05

Deployment in a perimeter

Install, upgrade and observe a system you cannot log into.

Upgraded without a remote session

Write about this layer
The answer, spoken
The runtime

Five layers, one deployment, inside someone else’s boundary

What we hold to02 / 04

Five things we hold to, and each one costs us something.

Not a culture statement. Each one is written so it could be shown false, because a value nobody outside the building can check is only a poster.

  1. Radical Ownership

    You take a problem end to end, including the deployment.

    Whoever picks up a layer designs it, ships it and watches it run inside a customer perimeter. Nothing is handed on at the edge of a job description.

  2. Engineering Excellence

    The hard parts are built here, not wrapped from elsewhere.

    Speech, turn taking, retrieval and the deployment are all ours to change. When a caller hears a pause, there is nobody else’s roadmap to wait on.

  3. Accelerated Growth

    You learn beside the people who built the layer.

    The team is small enough that a review comes from whoever wrote the thing you are changing. Mentoring here is an argument about your code, not a programme.

  4. Collaborative Culture

    An idea is judged on the call it would improve.

    Designs are argued in the open and the decision is written down with its reason. Disagreement is cheap; an unexamined assumption in a live deployment is not.

  5. Bleeding-Edge Tech

    A new model earns its place on measured audio.

    Speech and language move every month, so evaluation is the work: accents and code-mixed audio are measured per locale before anything reaches a caller.

What it ships into03 / 04

Your code answers calls that are already ringing.

The runtime is live in banking, health services and telecom, not in a demo environment.

A carrier’s support floor in daylight: two agents at desks with headsets, tall windows and a pale city skyline beyond.
In production

A carrier’s support floor on a live call

Illustrative scene, not a product capture

  • ~10,000 Voice AI agents live, 99.98% Agent Uptime, 30+ Languages, Eight industry-specific fine-tuned LLMs

    What the runtime carries today

    the runtime

  • 7 to 12 customer intents live in production, inside 6 weeks from go-live

    Intents added by the customer after go-live

    customer-led expansion, Jun 2026 to Aug 2026

  • 91K total calls since go-live, 38.9X volume growth, 67.54% AI automation rate

    Growth of one deployment since go-live

    all interaction types, Mar to Jun 2026

Open roles04 / 04

No roles are posted today.

That is the true state of it, and it does not mean we are not hiring. Write to the address in this panel with one paragraph about the hardest thing you have shipped, and a CV if you have one.

  • Tell us what you would want to work on, and we will tell you whether it exists here.
Write to the hiring team

hr@voicing.ai