The human in the loop. Cyber agent pilot.
Cybersecurity, AI & data science leader — 20+ years building production cyber-analytics platforms across enterprise, defense, and federal.
I build production-grade cyber data and analytics platforms — applying AI/ML, data science, and cloud engineering to defend networks and turn high-volume telemetry into decisions. Ph.D. in Computer Science, 20+ years across enterprise, defense, and federal environments.
Offline/air-gapped LLM fine-tuning, MCP & agentic AI workflows, and ML-based anomaly detection on live network telemetry.
Real-time situational awareness, threat detection at scale, and cyber-topology visualization for analysts and operators.
Large-scale analytics pipelines, cloud data engineering on AWS/GCP, and ML modeling over messy, high-volume data.
Part 2 adds a Wazuh SIEM on its own SOC VLAN, then reviews the whole build carefully, including a version pin meant to keep the agents in step with their manager that ended up upgrading all five past it.
cyberludusclaudecyber-rangeproxmox
Read more →Part 1 of a running log of building a cyber range by describing it to Claude Code connected to Ludus. Covers installing Ludus, deploying a segmented six-VM range from a paragraph of description, tapping its traffic passively, and generating realistic analyst activity for the tap to capture.
cyberludusclaudecyber-rangeproxmox
Read more →Introduces Breadcrumb, a small daemon-free tool that records where each opencode session lives, the git state of its workspace and a gist of the last prompt, so you can find and safely resume work across machines.
aiagentsopencodedeveloper-toolstypescript
Read more →Part 1 of a series on drone link security. Builds a simulated ArduPilot lab and the MAVIO workbench to measure what an unauthenticated MAVLink link exposes, from the parameter stream to commands that move the vehicle, and what that trust model means for the autonomy stacks, LLM-based ones included, that now sit on the other end.
cyberdronesmavlinkardupilotgazebo
Read more →Benchmarks an 8-bit Qwen3-Coder-Next running fully offline on AthenaBench's six threat-intel tasks. It scores 49.0 combined, the strongest open-weights result on the board and ahead of GPT-4 on CTI knowledge and ATT&CK technique extraction, and the post lays out a local retrieval plan for the two tasks where it struggles.
cyberaillmthreat-intelbenchmarking
Read more →