Collaboration
Worked with a 10+ person defense team to translate legacy radar processing into modern ML services that comply with security constraints.
Technical focus
- Built denoising autoencoders for clutter reduction and anomaly detection.
- Implemented pattern classifiers that flag maneuvers using sliding windows.
- Containerized pipelines with GitLab CI for reproducible simulations.
Outcome
The upgraded stack accelerated digital twin experiments and improved operator trust with clear metrics for recall/precision during test flights.