Enterprise risk thinking and production-grade AI architecture — the experience behind DeepMantis spans critical infrastructure risk management, multinational supply chain strategy, and the architectural decisions that let an autonomous agent run safely against real production environments.
Managing Partner
Hamburg, Germany
Architects the DeepMantis platform end-to-end — agent orchestration, sandboxing, multi-agent verification, and production-safety architecture. Started in data-driven retail-network analysis at Shell Germany. Multiple production AI systems shipped to market since pivoting full-time to AI.
LinkedIn
Managing Partner
Berlin, Germany
Oversees risk strategy and engagement methodology at DeepMantis — defining how scope is set, blast radius assessed, and regulatory exposure managed. Thirteen years across Shell, Toyota, and SABMiller: risk registers for refinery turnarounds, country-level supply chain strategy, and root-cause discipline on production operations.
LinkedInRisk discipline and engineering rigor — three operating commitments, enforced on every engagement.
No finding ships without a reproducible exploit. Browser-based verification, isolated per run.
Scope is signed and SHA-256 hashed before any tool runs. Out-of-scope is listed and stuck to.
Sandboxed execution. Scope-isolated agents. Safety enforced in architecture, not promised in policy.