AI/ML Engineer, Senior - WFH1659
$312,000–$416,000 year
RemoteUnited States
Job Summary
Design, build, and validate machine learning models for RF emitter identification using Python, PyTorch, and TensorFlow on constrained tactical edge hardware without cloud infrastructure. Conduct hands-on exploratory data analysis on NDF sensor datasets to characterize feature distributions, diagnose data quality issues, and produce documented findings. Implement and maintain ML data pipelines that ingest sensor streams, apply preprocessing logic, and ensure correctness on CPU-only Linux systems. Collaborate with technical leads to investigate sensor data reliability and validate assumptions under contention. Produce clear technical documentation of experiments and contribute to monthly status reports.
Required Qualifications
- Public Trust clearance
- US Citizenship
- 5–7 years of relevant experience
- BS or MS in Electrical Engineering, Computer Science, Applied Mathematics, or a closely related quantitative field
- 5+ years of hands-on applied experience in machine learning, data science, or RF signal processing
- Demonstrated proficiency in Python for ML and data science work — PyTorch or TensorFlow for model development, Pandas/NumPy for data manipulation, and scikit-learn or similar for evaluation and baseline modeling
- Hands-on experience designing, training, and evaluating deep learning models — particularly metric learning, Siamese networks, or other similarity-learning architectures — on real-world, noisy, imbalanced datasets
- Practical experience handling real-world data quality problems — missing values, label noise, class imbalance, systematic bias, and sensor artifacts — and the ability to diagnose and address them without discarding valid data
- Ability to develop and run ML pipelines on Linux-based systems without cloud infrastructure or GPU acceleration — optimizing for CPU-only inference and multi-threaded data processing on resource-constrained x86 hardware
Desired Qualifications
- Familiarity with RF signal characteristics, passive receiver phenomenology, and sensor data interpretation — including awareness of processing artifacts, attribution ambiguities, and measurement limits common in signals intelligence datasets
- Hands-on experience applying machine learning — particularly metric learning, deep learning networks, or similarity-learning architectures — to RF or time-series signal data, including feature engineering, training pipeline development, and model validation
- Exposure to TDMA network protocols or military datalink systems, and interest in learning the signal processing challenges of dense, contested electromagnetic environments
- Familiarity with direction-finding, time-difference-of-arrival (TDOA), or related passive geolocation concepts — understanding of their mathematical foundations and common failure modes is more important than operational experience
- Experience with binary serialization formats (FlatBuffers, Protocol Buffers) and high-throughput sensor data pipelines operating in near-real-time on resource-constrained hardware
- Background in statistical signal processing — error ellipses, bearing estimation uncertainty, feature reliability under noise — with the ability to distinguish statistically significant findings from artifacts of small sample size or improper normalization
- Relevant Certifications in machine learning, data science, or related technical fields (e.g., TensorFlow Developer Certificate; PyTorch Certified Associate; AWS Certified Machine Learning — Specialty; Microsoft Certified: Azure AI Engineer Associate; Certified Analytics Professional (CAP); etc.)
Hiring someone like this?
Get your role in front of qualified candidates on Sorce.