Testing AI agents where they break.

I'm Matheus, a senior AI security engineer working at Toloka, building RedThread —an autonomous AI red-teaming and defense-evidence engine for LLM and agentic systems.

Writing

Writing about AI security, red teaming, agentic systems, and building with LLMs:

• Constitutional Classifiers v2: What 1,736 Hours of Red Teaming Revealed About Production AI Defense
• The Red Teamer's Guide to Agentic Defenses: 12 Systems and Where They Break
• FlowSteer: The Attack That Current Multi-Agent Defenses Can't See

Connect