Senior Staff Audio Machine Learning Engineer

Boston Dynamics•Waltham, MA

About The Position

Boston Dynamics is seeking a Senior Staff Audio Machine Learning Engineer to join the Spot Sensing team. The role focuses on advancing Acoustic Change Detection (ACD), transforming beamformed and raw audio into actionable machinery prognostics. The engineer will also contribute to multimodal mechanical inspection on Spot by integrating acoustic features with image, video, and thermal data, and fusing multimodal representations with LLMs/VLMs to provide end users with prescriptive maintenance actions. The work uses a fleet-gathered industrial sensor dataset and Spot equipped with the Sorama L642 acoustic camera and SpotCAM2. The role owns acoustic machine-learning and digital signal-processing pipelines, improves anomaly detection and failure classification, and develops supervised multimodal pipelines for automated fault diagnosis. In collaboration with the broader Spot Apps organization, the engineer will combine acoustic insights with visual, thermal, and VLM/LLM data streams to create unified machine-health indicators, diagnoses, and corrective-action recommendations. The solution will be benchmarked, optimized, deployed, monitored, and continuously improved for industrial customers.

Requirements

At least 7 years of deep practical expertise in audio feature extraction, audio fingerprinting, and segmentation, and applying machine learning to beamformed and raw audio for anomaly detection.
Advanced proficiency in Python and C++.
Mastery of numerical and scientific computing tools such as NumPy, SciPy, scikit-learn, and PyTorch.
Expertise in time-domain and frequency-domain data analysis.
Expertise in multichannel array processing, including beamforming methods such as Delay-and-Sum, MVDR, and null-steering; Direction of Arrival estimation; and synthetic aperture techniques.
Expertise in unsupervised and supervised classification architectures for anomaly detection in real-world conditions.
Eligible to work in the United States; visa sponsorship is not available.

Nice To Haves

Experience in cross-modal fusion, combining audio representations with computer vision, thermal imaging, or multimodal foundation models/VLMs.
Background in rotating machinery health monitoring, including bearings, gearboxes, and pumps, and familiarity with mechanical vibration/acoustic failure standards.
Familiarity with TypeScript and modern web services for integrating work into Orbit, the fleet management software.

Responsibilities

Design, refine, and scale digital signal-processing and machine-learning pipelines for the Acoustic Change Detection system.
Improve and modernize the acoustic detection stack, including transitioning from unsupervised models such as BGMM to state-of-the-art anomaly-detection architectures.
Architect end-to-end supervised multimodal pipelines for automated fault diagnosis.
Integrate acoustic insights with visual, thermal, and VLM/LLM data streams to produce unified machine-health indicators, mechanical diagnoses, and recommended corrective actions.
Benchmark, optimize, ship, monitor, and continuously improve solutions for industrial customers.