Senior Machine Learning Engineer

Echodyne Corp•Kirkland, WA

3h•$124,733 - $187,048

About The Position

Echodyne is seeking a Senior Machine Learning Engineer to develop machine learning capabilities for its radars and data systems. The role works with radar data to improve labeling and enhance detection, classification, and tracking of targets in complex, changing environments. The engineer owns work from data exploration and model prototyping through evaluation and production deployment, making practical tradeoffs among accuracy, latency, computation, and reliability. The work supports advanced data systems and fielded radars used to protect infrastructure and people.

Requirements

Strong foundations in machine learning, probability, statistics, and experimental design.
Experience taking machine-learning solutions from prototype through production deployment.
Experience with time-series classification.
Experience training and evaluating models on imbalanced datasets.
Experience designing or assessing data augmentation for real-world sensor data.
Experience evaluating model performance, investigating failure modes, and improving data or models based on findings.
Experience deploying models on embedded, resource-constrained systems.
Strong Python software development skills and experience with TensorFlow.
Ability to independently lead an open-ended technical problem and communicate results clearly.
MS with at least 4 years, or PhD with at least 2 years, of relevant experience in computer science, engineering, applied mathematics, or a related field.
U.S. citizenship; the role may require handling export-controlled data.

Nice To Haves

Experience with radar, lidar, or other spatial sensor data.
Experience with real-time processing of streaming data.
C++ experience and familiarity with production software development practices.
Experience designing labeling workflows, managing large datasets, or using active learning to improve data quality.

Responsibilities

Prioritize real-world robustness and ensure zero-regression deployment of models to fielded radars.
Improve data-labeling accuracy and capacity by developing active learning and automated labeling systems and advising on tool design.
Design data augmentation methods that reflect realistic variation in radar deployment environments.
Improve existing track-classification models through hyperparameter tuning and evaluation of alternative architectures.
Research and develop new machine-learning applications, including self-supervised generative models using unlabeled data.
Build training and evaluation pipelines using field data, with attention to rare targets, imbalanced classes, and biased datasets.
Define evaluation metrics and assess model performance across target classes and field conditions, especially rare, high-impact events.
Analyze model failures, data gaps, and changes in operating conditions to improve robustness.
Partner with cross-functional teams to integrate models into production products.