Full-time position
Machine Learning Engineer
Neuralink
- South San Francisco, United States · On-site
- Machine Learning, Biomedical Engineering and
About
Neuralink's intracortical brain-computer interface decodes intent from spiking activity recorded by implanted electrode arrays. This machine learning role works on two open decoding problems: controlling a full virtual arm and hand with 29 degrees of freedom, and streaming prosodic speech synthesis in a user's own voice from intracortical signals. The work includes training models across sessions and participants on more than 50,000 hours of clinical-trial neural data so decoders hold up through weeks of neural drift without supervised recalibration, online adaptation for closed-loop co-adaptation between user and decoder, compressing decoders for on-implant latency budgets, and a pre-training, fine-tuning and reinforcement-learning stack for voice synthesis.
The posted base salary range is $129,000–$331,000 USD.
Requirements
Bachelor's degree or higher in computer science or a related field, or equivalent experience; experience designing, training and shipping real-time ML systems; strong sequence-modelling instincts (speech, robotics, reinforcement learning, time series or sensor fusion). Helpful: intracortical BCI decoding literature (handwriting and speech decoding, manifold alignment, unsupervised recalibration, cross-participant transfer), model compression or on-device inference. No neuroscience background is required. The role is on site five days a week.
Topics
- speech synthesis, search World Wide
- intracortical brain-computer interface, search World Wide
- spiking activity, search World Wide
- virtual arm and hand, search World Wide
- 29 degrees of freedom, search World Wide
- prosodic speech synthesis, search World Wide
- neural drift, search World Wide
- online adaptation, search World Wide
- closed-loop co-adaptation, search World Wide
- reinforcement-learning, search World Wide
Source: Neuralink