nyra labsnyralabs

Careers at nyra labs

Build the models that listen closer

We work on the details most systems remove, publish the models and evidence behind them, and put the strongest ideas into real clinical use.

See open roles

A lab with a real feedback loop

We research speech where the real signal matters

Our models preserve disfluencies, recover precise timing, separate the factors a voice carries, and learn from speech encountered outside the studio.

The work is published through open research and grounded in the therapy products of nyra health.

What you would help build

One program, from faithful measurement to speech understanding

  1. 01

    Preserve

    Capture what was actually said, including the parts clean transcripts remove.

  2. 02

    Measure

    Locate every word, sound, speaker, and event at the level where it occurs.

  3. 03

    Represent

    Separate content, speaker, prosody, and acoustics into interpretable spaces.

  4. 04

    Understand

    Build speech models that can learn from the whole signal.

How we work

Serious research, without the distance from reality

  1. 01

    Start from the signal

    We do not clean away the hard parts of speech. The pauses, repairs, overlaps, and vocal sounds are the research problem.

  2. 02

    Publish the proof

    A result should survive outside our lab. We release papers, open weights, datasets, and benchmarks that others can inspect and rerun.

  3. 03

    Close the loop

    Research reaches real therapy through nyra health. What fails in practice comes back to the lab as the next problem to solve.

  4. 04

    Own the problem

    The team is small by design. You define questions, build the system, evaluate it honestly, and communicate the result.

What you can expect

A small lab with room to do defining work

Research ownership

Work on a clear scientific problem with room to shape the method, evaluation, and release.

Work with your name on it

Contribute to papers, model releases, and benchmarks that make the work visible and useful.

Clinical grounding

Learn from speech technology used in therapy, where details matter and feedback is real.

Vienna, hybrid

Regular in-person collaboration at our home office, with focused time to think.

Join the lab

Open roles

  • Research

    Researcher, Post Training

    Shape how our speech models improve after pretraining. Work across feedback, data, evaluation, and training methods to make model behavior more accurate, controllable, and reliable.

    Vienna · hybridFull-time
    Apply
  • Applied research

    Applied Researcher, Audio

    Turn new audio research into models people can use. Prototype and scale work across speech understanding, representation learning, data, and evaluation, from first experiment to release.

    Vienna · hybridFull-time
    Apply
  • Platform

    Software Engineer, ML Platform

    Build the systems behind every experiment and release. Own parts of the data, training, evaluation, and inference platform that let a small research team move quickly and reproducibly.

    Vienna · hybridFull-time
    Apply

Open application

Your role is not listed?

Tell us which part of the research program you would move forward, what you would build first, and why you are the person to do it.

Write to us