I build defensive AI/ML systems, MLSecOps infrastructure, and resilient backend services. My work focuses on defense-in-depth LLM gateways, automated network intrusion detection, and reliable state machines. Currently engineering intelligent pipelines with LangGraph, PyTorch, and containerized FastAPI runtimes.
3-tier defense-in-depth gateway stopping prompt injection and PII leakage in LLM support, combining deterministic sanitization, regex scanning, Llama PromptGuard 2-86M, and LangGraph.
Network attack classifier trained on the ACI-IoT-2023 dataset to catch threats missed by signature rules, served via FastAPI with a containerized React SOC dashboard.