About
One-person lab. Public numbers.
I am Matt. I run frontier-class models on hardware I own, log what happens, and post benchmarks and failures with caveats — inspired by lab-note accounts that show their work.
Skrlr AI Lab leads with local inference benchmarks and agent experiments. Agentic security evals are a real second focus — sandboxed red-teaming, not hype.
Background: systems and security engineering — risk, social engineering, phishing defense, insider threat — now applied to AI agents with tools.
Contact: Matt@skrlr.com
Lab notes: @therealskrlr on X