Members-Only
Recent Talks & Demos are for members only
You must be an AI Tinkerers active member to view these talks and demos.
April 22, 2026
·
Montreal
Argus: Real-Time LLM Certification
See how Argus, a real-time LLM control layer, classifies prompts, monitors generation, and certifies output before release, demonstrating LLMs are not black boxes.
Overview
I built Argus, an infrastructure-level inference control layer for LLMs that classifies a prompt before the model call, applies a model-calibrated control envelope, certifies the generated behavior with observable metrics, and only releases the output after remediation and closure checks pass. In the demo, Argus shows the full chain end to end.
Links
Tech stack