Senior Voice AI Engineer, Evaluation and Reliability
We are currently looking for a Senior Voice AI Engineer, Evaluation and Reliability to join a professional German-based product company team that designs innovative eHealth solutions for hospitals.
You will work on a stable healthcare product.
Senior level, and we mean it. You would own the evidence that this system is safe to put in front of
patients, on your own judgement, and tell us when the answer is no. That isn't a job we can hand to
someone still learning what to measure.
You would not be building a voice stack. We use a third-party voice platform that already handles the
telephony, the speech recognition and synthesis, and the knobs for end-of-speech detection and
interruption. What it does not do is tell us whether any of that is working well enough for patients, and it
gives us no way to check that a change hasn't made things worse. That's the job.
So you'd own the conversation and the evidence about it: how the assistant behaves when a caller
interrupts, corrects themselves or won't answer; how fast the whole thing responds once a round trip to
our backend is in the loop; and the body of test cases we run before every change.
You need
To have operated a phone or voice assistant that answered real calls from real users. You can tell us its response time and how you measured it. Give the response time as a percentile, the figure that most callers experienced or worse, not as an average. It is acceptable if you built the audio pipeline yourself, but that is not the important part. We want to know that you know what to measure and what good results look like.
To have tuned a voice system, not only delivered one: end-of-speech detection, barge-in, and how long to wait before the system assumes that the caller has finished. You would do this first with a vendor's settings, and later in our own platform. You would do it on phone audio, where clean barge-in is difficult.
To know what happens to speech recognition accuracy when the audio comes through a phone line and not through a good microphone. You have measured this yourself. You did not only read about it.
A way to test conversation changes that does not depend on listening to a few calls and hoping that they are good. This is the centre of the role. It can be a replay harness, a scored scenario set or another form. We want to hear how it worked and where it was not good enough.
Enough Python to build that harness and the tools around it.
Nice to have
To have built streaming audio infrastructure, a turn-taking system or a speech pipeline yourself. It helps with the half of the role that builds the platform.
Experience with a closed voice platform and the problems that it brings, not only with your own system.
Telephony background: SIP, WebRTC, Asterisk, Twilio or similar. It helps you understand why calls sound the way they do, but you would not operate any of it.
Work with health data, or in another field where mistakes have consequences.
German, but we do not depend on it.
You do not need
To speak German.
To know PHP or our existing backend. Another team builds the endpoints that you would call.
To have trained a model from the beginning, without a pre-trained model as the base.