Principal Applied AI Systems Engineer · Open to principal/staff roles · Remote from Southeast Georgia
I turn operational friction into governed AI systems: persistent agent orchestration, evaluation, enterprise integration, approval boundaries, deployment, and production operations.
Vice President, Integration Engineering
Senior integration engineering in a regulated fintech environment: business technology, enterprise integrations, identity-adjacent workflows, platform reliability, automation opportunities, and the ownership/documentation work that keeps complex systems supportable.
Founder & Principal AI Systems Engineer
Build and operate applied-AI, agentic LIMS, and managed-infrastructure systems: supervised specialist execution, evidence gates, AWS/Docker delivery, rollback, observability, reporting workflows, and recurring production support.
Four systems that show the whole job: agent orchestration and evaluation, operational AI delivery, distributed product engineering, and security-sensitive native architecture.
Governed agent workflows turn requirements, implementation, testing, release evidence, and operational follow-through into one bounded delivery system—compressing complex work from months to days without removing human approval.
$ friday run --bounded
context: persistent
specialists: routed
proof gates: enforced
A persistent AI operating system that routes work to fresh specialists, preserves context, enforces approval boundaries, evaluates evidence, and coordinates hundreds of meaningful executions and checks in an ordinary week.
$ blackglass status
envelopes: encrypted
devices: signed
replay: rejected
An encrypted iOS messaging system with multi-device identity, signed device frames, APNs wake-ups, replay protection, revocation, rollback journals, and terminal-failure quarantine. Operated across 3 users and 5 devices for 6 months.
A self-hostable cross-platform communications product spanning web, native iOS, Electron, messaging, file delivery, LiveKit voice, screen sharing, notifications, administrative analytics, and production operations.
How validation maps, parallel delivery lanes, native-workflow fit, production-like clone proof, and explicit human authority compressed LIMS delivery from months to days.
Automation becomes predictable when permission is part of the interface, not an assumption hidden behind it.
Decision: routine workloads will not consume the space needed to preserve, copy, and compare a failed state. Capacity planning will include a diagnostic maneuver, not just normal demand and restart overhead.
What I built, what broke, how I evaluated it, and the evidence that changed the next decision.
Read the blog →Best fit: principal/staff IC roles where applied AI, agentic platforms, evaluation, enterprise integration, and production operations must become one reliable system.
I'm targeting hands-on principal/staff roles building, evaluating, deploying, governing, and scaling AI systems in real business contexts—remote from Southeast Georgia.