At RAM Labs I worked across tactical cybersecurity, code vulnerability modeling, distributed edge inference, and cognitive security.
Core role
Member of Technical Staff from April 2022, then Senior Member of Technical Staff and principal investigator from April 2024 to March 2025. Across programs I owned modeling, data pipelines, evaluation, and integration, and wrote more than fifteen SBIR proposals.
Program snapshots
- T-UEBA: tactical UEBA for Zero Trust conditions using temporal-heterogeneous behavior modeling and adaptive risk calibration.
- DL-PATCHER: automated vulnerability repair workflows over code corpora with transformer-family modeling and curation-heavy fine-tuning.
- SpHyRE-Net: low-SWaP anomaly-focused modeling for constrained edge environments.
- DAICON: distributed AI execution over heterogeneous tactical nodes with resilience under ad hoc networking constraints.
- DEVIS: embedded vulnerability detection in binaries with ML-assisted path prioritization.
Evaluation
These programs required testing model behavior under changing data, limited compute, and unreliable connectivity.
Patent context
Co-inventor: Automated Bug Fixing Using Deep Learning Including Pre-training and Fine-tuning, U.S. application 18/375,839 (Geddes, Mabie, McGraw).