About
Builder of production agent systems for security and operations.
I build production AI agent systems. I focus on moving agents from "chat interfaces" to "infrastructure layers." At my last company, agents I built handled incident resolution and fleet deployment across hundreds of endpoints daily.
Track Record
What I've Built
-
End-to-end incident resolution
Agents that take hundreds of open security violations and turn them into actionable remediation: grouping by host and severity, drafting policy rules at multiple restrictiveness levels, and generating audit bundles for human review. Operators make the final call; the system handles everything up to that point.
-
Conversational fleet deployment
An installer agent that replaces multi-page runbooks with a guided conversation. It discovers the target environment, runs preflight checks, flags blocking conditions like insufficient disk, deploys asynchronously across the fleet, and reports back when hosts are live.
-
Diagnostic operations layer
A natural language interface over execution logs, configurations, and remote infrastructure. Teams trace failures to root causes, cross-reference settings across hosts, and surface mismatches without writing queries or SSH-ing into boxes.
What I Believe
-
Agents should own the activity layer.
Most workflows were designed around human bandwidth constraints, not actual needs. The repetitive execution, the data gathering, the cross-referencing. Humans show up for judgment and relationship moments. The question isn't optimization. It's elimination.
What I'm Building Next
-
Agents that own business outcomes.
Building systems where agents own entire business functions, not just individual tasks. The goal isn't automation for its own sake, it's outcomes.
Want to build something together?
Connect on LinkedIn