Talk To Me About

I’m spending a lot of my time thinking about what happens when AI stops being a demo and starts running real parts of the business. The interesting problems tend to show up in a few places:

Governing AI Agents in the Real World

As systems move from copilots to autonomous agents, security and control start to break down. I’m interested in the infrastructure that defines what agents are allowed to do, who they can interact with, and how risk is enforced while they are operating.

That includes:

• Intent-aware data security built for AI workflows
• Observability and audit trails for autonomous systems
• Identity, access, and runtime controls for both humans and agents

Making Inference Work at Scale

AI is now a production workload, and cost, latency, and reliability matter. I like teams building software that helps companies run and route inference efficiently across environments.

I’m especially drawn to:

• Edge and hybrid inference infrastructure
• Intelligent routing across models and compute environments
• Software optimized for specific hardware and accelerators

Resilient Physical and Industrial Infrastructure

Data centers and operational systems have quietly become mission-critical, yet the software that operates and protects them remains fragmented. I’m drawn to founders treating resilience as a software problem.

This often looks like:

• Security and access control for data center and OT environments
• Infrastructure observability that connects physical and digital systems
• Software that aligns workloads with power, thermal, and reliability constraints