Machine Learning Engineer, Radar Perception II/III

BLUE ORIGIN•Reston, VA

1d•$119,814 - $182,988

About The Position

Blue Origin’s Blue National Security (BNS) organization develops launch and spacecraft systems for national security missions. Within Autonomy & Intelligent Mission Products, this individual-contributor role focuses on AI-enabled products for demanding defense applications. The engineer will own the AI layer that processes raw radar data to detect and characterize objects, helping an autonomy stack or operator make decisions. This is a new development program with an opportunity to define the model architecture, working with limited labeled data and constrained hardware.

Requirements

Bachelor’s degree in Computer Science, Electrical Engineering, Computer Engineering, Physics, or a related field.
About 3 years of ML engineering experience.
Track record of building and shipping deep learning models in PyTorch or a similar framework.
Experience with edge deployment, such as TensorRT, quantization, or FPGA or SoC targets.
Strong Python skills.
Working knowledge of radar fundamentals, including range, Doppler, angle, resolution, clutter, and multipath.
Ability to obtain a TS/SCI clearance.
Must be a U.S. citizen or national, U.S. permanent resident, or lawfully admitted into the U.S. as a refugee or granted asylum.

Nice To Haves

Familiarity with Ground Moving Target Indicator (GMTI) or Airborne Moving Target Indicator (AMTI) signal-processing concepts and their implications for system design.
Experience with sensor data from radar, sonar, SAR, lidar, or RF.
Experience training on synthetic or simulated data and closing the gap between simulation and real data.
Experience in uncertainty quantification, calibration, or out-of-distribution detection.
Experience with embedded C++.
Active TS/SCI clearance.

Responsibilities

Design, train, and evaluate deep learning models, including GNNs, transformers, DRL, Learned Algorithm Unfolding, or hybrid approaches, to detect and classify objects in raw radar data and range-Doppler and range-angle maps.
Build models to estimate object attributes such as size, shape, motion state, track, and behavior, with calibrated uncertainty.
Represent radar data appropriately for each model, such as using point or track graphs for GNNs.
Build the data engine, including synthetic data generation, simulation-based training, labeling strategy, augmentation, and class-imbalance handling.
Detect inputs outside the training data and flag low-confidence results for downstream systems.
Work with Advanced Systems, DSP, GNC, and flight software teams to define data handoffs and deploy models on embedded or edge hardware.
Define metrics, test sets, and evidence needed to qualify ML models for edge deployment.