Building open-source LLM-security tooling. Interested in prompt injection, MCP rug-pull / shadowing, agent runtime invariants. Background in enterprise AI adoption and LLM/agent systems engineering.