Machine Learning Researcher

ML Research Scientist (Health & Sensing)

San Francisco, CA (Hybrid)

Up to $350K + Equity

Visa transfer possible

My client is the first sleep fitness company. 1 in 3 people are sleep deprived, and the CDC calls insufficient sleep a public health epidemic. Our mission is to fix this and fuel human potential through optimal sleep. They've raised $90M from leading Silicon Valley investors and are working on exciting new products.

They have been consistently recognized as one of Fast Company’s Most Innovative Companies, including the top wellness company for 2023.

About This Role

We is taking sensing beyond sleep into a new modality, and you'll build the algorithms for it from scratch: high-frequency biosignals turned into metrics that change behavior, trained on over a billion hours of sleep data. The AI/ML team went from 2 to 11 people in ten months under our VP of AI/ML, ex-Apple. This is 0 to 1, not maintenance.

If you've spent your career on raw sensor data and wanted a product that ships to hundreds of thousands of people, you'll feel at home here.

• Build 0 to 1. Train models for a sensing modality that isn't in the product yet.
• Work the raw signal. High-frequency motion and activity time series from IMU, PPG, and optical sensors.
• Model at scale. Transfer learning and self-supervised pretraining on a billion-plus hours of physiological data.
• Ship it. Get models onto real beds with a cross-functional R&D and production team.
• Own the metric. Define a new health signal and prove it holds against physiology.

Requirements:

• Has trained ML models on raw physiological signals measured off a human body. PPG, ECG, EEG, respiration, temperature, body- worn IMU. Research or industry both count.
• Python and a deep learning framework, used to train models rather than call APIs. PyTorch or TensorFlow.
• 3 - 10 years of experience in applied ML on physiological or motion sensor data, using Python
• Signal processing depth on high- frequency time series: filtering, feature extraction, artifact rejection on raw sensor output.
• Large- scale physiological modeling with transfer learning or self- supervised pretraining.
• Graduate degree in biomedical engineering, bioengineering, physiology, or electrical engineering, on top of ML depth.

Why This Role Stands Out

A sensing modality that doesn't exist yet: next-generation consumer health hardware, a new modality beyond sleep, built 0 to 1. Not tuning an existing pipeline.

Cutting-Edge AI

Named projects: a multimodal health foundation model, a closed-loop RL thermoregulation system, and a physiological simulator for healthy aging. The sensing people you'd want are already there. The VP of R&D ran health monitoring devices at Verily and led Alphabet's first FDA-cleared wearable.

Key Highlight

The sensors architect has 10+ years in wearables. For this pool that beats the brand.