Joe Krisciunas
adversarial researcher · systems engineer
I build systems and break AI. After stints at Collins Aerospace, TIBCO, AWS, and Julius AI (founding engineer, 3M+ messages/week), I now research adversarial vulnerabilities in agentic LLM pipelines under the Cartograph banner.
My work maps the attack surface of AI agents — from prompt injection to multi-agent escalation — and develops principled defenses through structured empirical testing.
Career
Avionics software engineering, embedded systems for flight-critical applications.
Backend systems and data integration platform development.
Large-scale distributed systems and cloud infrastructure.
Built the core AI data analysis platform from the ground up. 3M+ messages/week.
Developer tooling — TypeScript SDK and deployment CLI for the Whop platform.
Independent research into AI system vulnerabilities and agentic threat models.