Machine Learning Research Scientist/Engineer, Audio

Scale partners with the world's leading AI labs to accelerate progress in frontier generative AI. We are building a dedicated research team focused on pushing the boundaries of speech generation, speech recognition, and speech-to-speech transformation. We are hiring Research Scientists and Research Engineers with deep expertise across text-to-speech (TTS), speech-to-speech (STS), and automatic speech recognition (ASR) to help define the next era of human-AI communication.

In this role, you will invent and deploy new algorithms to expand the capabilities, fidelity, and generalization of large-scale audio models. You will shape research directions at the intersection of modeling, data, and evaluation for speech systems. You will collaborate with world-class researchers and directly contribute technical and strategic insights to the development of the next generation of open and proprietary foundation models in audio. We encourage collaborations within the industry and academia, and support the publication of research findings. Successful candidates will partner with top foundation model labs, providing both technical and strategic input on the development of the next generation of generative AI models.

You will:

• Develop novel training and post-training techniques for audio models, advancing core areas such as speaker adaptation, prosody control, noise robustness, etc.
• Design new reward models and preference optimization techniques specifically for speech quality, emotion, and intelligibility in TTS, STS, and ASR.
• Investigate model failure modes in realistic deployment settings and propose scalable solutions for bias mitigation, robustness to accents, and long-tail speaker variation.
• Create industry-standard evaluations for the world’s leading speech models 
• Publish breakthrough research at leading conferences and shape best practices across the speech research community.
  
  

We’re looking for:

• Ph.D. or Master’s degree in Computer Science, Machine Learning, Electrical Engineering, or a related field with a focus on speech or audio processing
• Deep understanding of neural generative models for audio, including recent advances in transformer-based architectures, diffusion models, and large-scale pretraining for speech.
• Experience with fine-tuning, reinforcement learning, or reward modeling for audio quality optimization or speech preference modeling.
• A strong research portfolio with publications at top venues such as Interspeech, NeurIPS, ICLR, ICML, ACL, or similar.
• Exceptional written and verbal communication skills, with the ability to clearly convey complex technical ideas to both internal and external partners.
• Prior experience working directly with model training, data curation, or evaluation at scale is a significant plus.