Research & Innovation

Applying supervised and unsupervised ML models to identify adversarial behavior patterns in large-scale network telemetry — helping accelerate threat detection and response.

Fine-tuning and deploying open-weight large language models on government-controlled infrastructure to enable secure, air-gapped NLP capabilities without dependence on commercial APIs.

Developing XAI frameworks that make ML model outputs auditable and interpretable for federal decision-makers — meeting emerging OMB AI governance requirements.

Evaluating and implementing NIST-standardized post-quantum algorithms (CRYSTALS-Kyber, CRYSTALS-Dilithium) for federal PKI infrastructure migration planning.

Policy decision engine prototypes that dynamically adapt micro-perimeter rules based on real-time device health scores, user behavior baselines, and threat intelligence feeds.

Developing detection and mitigation techniques for adversarial attacks against AI/ML systems deployed in federal environments, including data poisoning and model inversion attacks.

Architecting lightweight security control frameworks for tactical edge devices in DIL environments — enabling mission continuity without centralized infrastructure.

Applying reinforcement learning and constraint satisfaction models to autonomous mission planning problems where real-time human oversight is limited.

Security protocols for military mesh networking environments, including key management, authentication, and anomaly detection in bandwidth-constrained tactical networks.

Developing probabilistic fusion models that integrate structured and unstructured data from heterogeneous sources — improving intelligence completeness and reducing false positive rates.

Applying time-series forecasting and causal inference to defense logistics datasets, enabling predictive maintenance, supply chain optimization, and readiness forecasting.

Computer vision and NLP pipelines that automate extraction and analysis of geospatial intelligence from satellite imagery and open-source reporting streams.