Igor Tarpan
AI Security Engineer — Agentic Systems / Threat Intelligence
AI-native security — multi-agent threat intel, IR triage, MCP/tool security, running in production. Lately, more platform than tool: something other security engineers build on, fed by a shared context layer instead of everyone re-deriving the same investigation context. Security Engineer at Superhuman, previously Amazon.
Writing about where LLM systems and security operations actually meet — the operational version, not the hot-take version.
writing
featured
talks
- Incident Response Triage with LangGraph — LangChain NYC Meetup
- Best Practices for Securing AI Agents — Moldova
- Predict — Recorded Future, London
- Wizdom — Recorded Future, NYC