Cantina Labs
Machine Learning Engineer, TTS
Europe +1 · Remote · Full-time · Posted Apr 29, 2026
- Location
- Europe
United States - Experience
- Exceptional research/development experience with large-scale audio models (>3B models and >500k hours data); no years specified.
About the job
Build and productionize large-scale text-to-speech and related speech models, from data and experimentation through evaluation and inference.
What you'll do
- Architect, train, fine-tune, and align large-scale speech models; lead research projects and design experiments.
- Own data requirements and develop objective and subjective model evaluations, including robustness, bias, and safety testing.
- Harden and scale training, evaluation, and inference pipelines; optimize latency, memory, and cost for production.
What they ask for
- Exceptional research/development experience with large-scale audio models (>3B models and >500k hours data).
- Hands-on experience with transformer architectures and/or diffusion models and/or audio language modelling; strong multi-node and multi-GPU distributed training experience.
- Strong software engineering skills, including PyTorch and performance work (profiling, CUDA/Triton/C++ as needed); able to write reliable production-quality code.
- Shipped large-scale speech/audio models to production; background working with large-scale ML data.
Nice to have
- Experience with voice-cloning, speech-control, or voice-generation.
- Notable publications and/or open-source contributions in speech/audio/ML.
Tools
PyTorch, CUDA, Triton, C++, GRPO, DPO, SV/WER/ASR
AI focus
Develop safe, steerable, and trustworthy TTS and related speech systems, including voice cloning, controllable TTS, voice conversion, evaluation, and safety and misuse mitigation.
Model-written summary · the original posting is authoritative