Who
I'm Jhon Arango — eight years of hands-on experience and research in data engineering, machine learning, and GPU hardware setup and configuration; with frontier models as a force multiplier, that depth is what let a single engineer build and operate ARKONA, a production agentic AI platform on on-premise GPUs — on a 20+ year foundation of cybersecurity systems engineering across the U.S. Air Force, NSA, and defense industry. CNODP graduate. Chief Cybersecurity Engineer
at NSA. Founder of Arkona Research, where I design and ship production
agentic AI systems with defense-grade rigor.
Philosophy
Agents are systems. Treat them like safety-critical infrastructure: failure modes first,
observability built in, human override at every level, and an audit trail that survives
the next regulator visit. Speed comes from discipline, not from skipping it.
Core competencies
- Agentic systems — Claude Code SDK, MCP, multi-agent orchestration, tool harness design.
- AI governance — NIST AI RMF, ISO 42001, FAIR, RACI, OSCAL; COMET delegation framework.
- Security engineering — SIGINT, OT/ICS reverse engineering, MITRE ATT&CK for ICS, threat modeling.
- Hybrid AI infrastructure — local LLMs (Llama/Phi/Gemma), dual-GPU routing, cost-aware cloud fallback.
- Voice + multimodal — CIPHER real-time voice stack (Whisper + Kokoro + 76 tools).
What I'm looking for
Teams building agentic systems where reliability, governance, and production discipline
actually matter — not demos. Particularly interested in Anthropic's Agentic Systems
research track. Starting a D.Eng. in AI/ML at GWU in Fall 2026 alongside the work.