Founding Research Scientist

About the role

Miso Labs aims to achieve the voice Turing test by moving beyond the current STT->LLM->TTS paradigm and training full-duplex speech-to-speech models. This involves exploring new architectural ideas and scaling them rapidly.

Responsibilities

• Explore and develop new architectural ideas for speech-to-speech models
• Contribute to the entire model training stack, from small-scale experiments to large-scale pre-training runs on up to 1000 H100s
• Work in an in-person culture in San Francisco, with relocation assistance provided
• Earn recognition for your work and have the opportunity to publish your findings at Miso Labs in conferences and on open-source platforms
• Sponsor travel to top AI/ML conferences

Requirements

• Preferred Qualifications:
• Have a Bachelor’s Degree in a technical field (Computer Science, Physics, Mathematics etc.) or equivalent experience
• Excited to work on all aspects of model training and development
• Comfortable with both research and hands-on engineering
• Capable of owning long-term research outcomes end-to-end
• Willing to learn technical skills quickly, in a self-directed manner
• Prior experience with both PyTorch and CUDA

Qualifications

• Exceptional ability in technical fields such as Computer Science, Physics, or Mathematics
• Ability to contribute to the entire model training stack, from small-scale experiments to large-scale pre-training runs on up to 1000 H100s
• Self-motivated and capable of working independently
• Open to learning new technical skills quickly

Benefits

• Full benefits including healthcare, vision, and dental
• Relocation bonus