Ideal candidate:
- Authored or contributed to key generative audio projects (WaveNet, AudioLM, Tacotron, Tortoise, AudioCraft, MusicGen, fairseq, wav2vec, etc.)
- Experience training and implementing inference for transformers and diffusion models
- Proficient in Python
- Proficient in at least one of PyTorch, Tensorflow, or JAX
Technologies:
- PyTorch, TorchAudio
- Numpy/Scipy, Jupyter, Pandas
- CUDA / C++
Preferred:
- CUDA performance nut
- MLOps experience (Ray, KServe, Triton, Seldon)
- Experience with Kubernetes
- Experience designing distributed system architectures able to handle high traffic and concurrency
- Audio engineering background