Engineering Roles

You’ll tackle machine learning problems that most tech companies never encounter. Your data comes with constraints that rarely exist elsewhere at scale: you’re working in cleared networks, against adaptive adversaries, often with signal and imagery that carry classification markings. The challenge is not just building a model. It is building one that retrains fast enough to track a changing threat, compresses enough to run at the edge, and integrates into systems where latency and reliability are non-negotiable.
You’ll develop training pipelines that ingest diverse data streams—RF, imagery, geospatial, temporal, unstructured text, graph-structured intelligence—and tailor them for models that perform detection, classification, anomaly detection, forecasting, and decision support. On the imagery side, that means detection and change detection on overhead and airborne collection, where targets are small, look angle and season change the scene, and a model trained on one sensor does not automatically work on the next. You’ll implement those models in PyTorch or TensorFlow and optimize them for inference in constrained compute environments, whether that means cleared cloud, edge devices, or embedded platforms. Sometimes you’ll publish and defend your work through peer review. Sometimes your work will have to remain classified, because it matters.