Applied Machine Learning Scientist

OmniBuds; Cambridge; Hybrid; GBP 70,000 - 110,000; Full time

Job summary

OmniBuds is seeking an Applied ML Scientist in Cambridge to turn multimodal, noisy wearable signals into reliable physiological insights. You will work at the intersection of ML, signal processing, and physiological sensing for ear-worn devices.

You will lead projects from signal characterisation to prospective evaluation, collaborating with clinical science, sensing, hardware, and firmware teams to deploy efficient algorithms for real-world use.

Qualifications

• Strong background in applied machine learning and signal processing.
• Experience with multimodal, physiological or biomedical sensor data.
• Proficiency in Python and modern ML frameworks.
• Good understanding of model validation, uncertainty, bias and robustness.
• Experience with wearable/edge ML and resource-constrained inference is a plus.
• Ability to work across ML, clinical science, sensing, hardware and engineering.

Responsibilities

• Develop signal-processing and machine-learning methods for multimodal in-ear physiological signals, including PPG, cardiovascular acoustics, IMU, and temperature.
• Build models that fuse complementary sensing modalities to infer cardiovascular and autonomic physiology from noisy, incomplete data.
• Develop methods for signal-quality assessment, measurement opportunity detection, uncertainty estimation, calibration, and model abstention.
• Lead the scientific pipeline from signal characterisation and representation learning through model development, validation, and prospective evaluation.
• Investigate multimodal fusion, temporal modelling, representation learning, and hybrid physiological + ML approaches.
• Design rigorous experiments and ablation studies to understand information use and reliability of inferences.
• Collaborate with clinical data scientists, clinicians, hardware, and firmware teams to optimize the sensing and inference system.
• Translate research models into efficient algorithms suitable for real-world and on-device deployment.
• Contribute to OmniBuds’ roadmap for blood-pressure estimation and longitudinal cardiovascular health.

Skills

Applied ML

Signal processing

Time-series modelling

Computational physiology

Python

Model validation

Uncertainty estimation

Robustness

Cross-disciplinary collaboration

Tools

Python (ML frameworks)

Job description

Location

Cambridge, UK

Department

Research, Algorithms & Engineering

Employment Type

Full-time | Hybrid

About The Role

As an Applied ML Scientist, you will develop machine-learning methods that turn multimodal, noisy wearable signals into reliable physiological and clinical insights. You will work at the intersection of machine learning, signal processing, and physiological sensing, developing models that can recover cardiovascular and autonomic information from real-world data and ultimately operate within the constraints of ear-worn devices.

What you’ll do

• Develop signal-processing and machine-learning methods for multimodal in-ear physiological signals, including PPG, cardiovascular acoustics, IMU, and temperature.
• Build models that fuse complementary sensing modalities to infer cardiovascular and autonomic physiologyfrom noisy, incomplete, and context-dependent data.
• Develop methods for signal-quality assessment, measurement opportunity detection, uncertainty estimation, calibration, and model abstention.
• Lead the scientific pipeline from signal characterisation and representation learning through model development, validation, and prospective evaluation.
• Investigate multimodal fusion, temporal modelling, representation learning, and hybrid physiological + ML approaches.
• Design rigorous experiments and ablation studies to understand what information models are using, when inference is reliable, and where it fails.
• Work closely with clinical data scientists, clinicians, hardware, and firmware teams to optimise the complete sensing and inference system.
• Translate research models into efficient algorithms suitable for real-world and ultimately on-device deployment.
• Contribute to OmniBuds’ scientific roadmap for blood-pressure estimation, hypertension, and longitudinal cardiovascular health.

What we expect

• Strong background in applied machine learning, signal processing, time-series modelling, or computational physiology.
• Experience developing ML methods for multimodal, physiological, biomedical, or other complex sensor data.
• Strong proficiency in Python and modern ML frameworks.
• Good understanding of model validation, generalisation, uncertainty, bias, and robustness.
• Experience with wearable/edge ML or resource-constrained inference is valuable, but not essential.
• Ability to go beyond maximising model accuracy and ask the harder scientific questions: why does the model work, when does it work, and when should we not trust it?
• Ability to collaborate across ML, clinical science, sensing, hardware, and engineering.

Why OmniBuds

You’ll work on problems few teams in the world are tackling—bringing continuous, medical-grade inference onto tiny devices worn all day, every day.