Founding Machine Learning Engineer
full time · remote or on-site · founding equity
Own Spex-TTS end to end — the model, the serving stack, and the 107 ms budget it answers inside.
You will own
- Voice models that hold one identity across 23 languages
- The GPU serving fleet: batching, warm spill, failover
- First audio inside the human handoff, at production load
- Publishing what you measure — the site quotes the engine, not a roadmap
What we look for
- You have trained or served speech or audio models in production
- You measure before you claim
- You want the model and the metal in the same job
