devlocal
PrivateA self-hosted engineering shop staffed by AI engineers. Each agent has a durable identity; each task runs in a disposable sandbox with credentials that expire with the run.
// principal engineer + security architect + agentic systems
23+ years building distributed systems, security platforms, agentic infrastructure, and the tooling that keeps critical systems running.
My career spans the U.S. Army, IBM, nearly 15 years at GoDaddy, and most recently InspectionGo, where I led security architecture and agentic engineering. Along the way I've worked in signal operations, systems administration, systems security administration, networking, site reliability engineering, software engineering, and security engineering. The constant was designing and operating complex production-critical systems at scale, then building the tooling and guardrails that made them safer, more resilient, and easier to run.
The projects below reflect that mix: security that's structural, not bolted on, infrastructure that earns its uptime, and agentic systems that are governed instead of improvised.
// core disciplines
// languages / ecosystems
A self-hosted engineering shop staffed by AI engineers. Each agent has a durable identity; each task runs in a disposable sandbox with credentials that expire with the run.
Source-first programming language and runtime contract. Explicit effects, capability-typed authority, supervised actors, post-quantum-aware policy, and runtime evidence — by design.
Self-hosted ITSM, ITIL and CMDB in one binary. The CMDB is the spine and everything else is a query over it: tickets, workflows, AI steps behind human approval, and a hash-chained audit trail.
Secure-by-default BEAM-native AI workspace. LiveView dashboard, multi-backend chat, plug extensions, and BeamAgent-backed sessions.
BEAM-native SDKs for multi-backend agents. Unified sessions, routing, policy, memory, audit, and capability parity across runtimes.
Optimization-driven local inference on tiny VRAM. Built to fit 120B to 160B-class models onto 8GB hardware without quantizing away fidelity.