Manager, Machine Learning on Orbit

Boston Dynamics•Waltham, MA

1d•$187,580 - $275,117

About The Position

Boston Dynamics is seeking a highly experienced and motivated Manager, Machine Learning to support their Orbit team. The Orbit team is building a foundation model with leading zero-shot performance on common industrial inspection tasks, leveraging millions of multimodal data points from industrial environments captured by the Spot quadruped robot. This role offers the opportunity to design, build, and continuously improve Orbit’s machine learning systems. The position is within the Spot Applications Organization, working alongside full stack engineers, product management, and design, with support from ML platform engineers and collaboration with researchers across Boston Dynamics.

Requirements

BS and 10+ years of experience, or MS and 7+ years of industry experience in Computer Science, Machine Learning, Robotics, or a related field.
Expert-level software development skills, particularly in python, with a strong command of software architecture, design patterns, and performance optimization techniques.
Experience training, fine tuning, and applying state of the art generative ai models. Preferably focused on computer vision tasks.
Experience with common ML tools (CUDA, pytorch), architectures and workflows.
Ability to track the current state of the art, and identify opportunities for product applications.
A willingness to dive deep into challenging problems in computer vision, develop creative customer-focused solutions in collaboration with teammates, and clearly communicate findings.

Responsibilities

Be a lead on a team of ~5 machine learning engineers.
Research, prototype, and deploy models leveraging multimodal data from robotic sensors to extract useful site information - anomalous machine states, object classification, etc.
Establish effective methods of evaluating inspection model performance.
Curate datasets to be used in model training, fine-tuning and evaluation.
Identify and employ model performance improvement strategies - e.g. fine tuning, model selection, and prompt engineering.