I am an AI practitioner-researcher based between London and Berlin. My work focuses on reliability and safety in agentic systems.
Previously at Gradient Labs and QuantCo. I combine engineering rigor with a background in computational social science to build systems that work in the real world, not just benchmarks.
.svg)
Flank
In productionAgentic legal workflows for law firms and in-house teams
- Lead AI strategy and product roadmap for autonomous legal agents
- Redesigned document review and objection handling, achieving 96% accuracy in identifying objections and flagging provisions requiring human judgment
- Built fully autonomous agent handling multi-step protocols, direct counterparty communication, obstacle navigation (e.g., OOO replies), and boundary enforcement
- Design evaluation systems for accuracy, auditability, and alignment with professional standards
Sentra
PilotAI mental health companion for veterans and first responders
- Co-founded and built first prototype for AI-powered mental health companion in crisis-adjacent populations
- Pilot results: 82% of participants rated responses as highly relevant; 64% reported immediate stress relief; zero cases of increased anxiety
- Implemented suicide intent detection with immediate human handoff protocol
- Developed fully personalized conversational system with user-based knowledge graph guided by latest PTSD research
- Partner with Wounded Warrior Project and UK Ministry of Defence
PsySafe
Open sourceOpen toolkit for mental health crisis response in chatbots
- Risk detection patterns and escalation protocols for mental health crises
- Policy templates for harm reduction and refusal behaviors
- Ethics-based initiative exploring LLM-based detectors for mental health risk and ethical response pathways
- Used by practitioners building crisis-sensitive AI systems
