Nediem Khalfallah

I’m a Software Engineer building AI-driven security systems—from production-grade intrusion detection to LLM guardrails. I focus on turning threat detection into reliable pipelines that stand up in real environments, with strong attention to architecture and deployment.

In my capstone project, Pulse, I helped deliver a hybrid intrusion detection and prevention system that flagged 140+ security alerts across 2 months in production and aimed to cut mean time to detect by an estimated 70%. I implemented stacked ensemble models (Random Forest, XGBoost, TabNet, LightGBM) for a two-stage NIDS/HIDS architecture over MQTT, with a multi-tenant NestJS/PostgreSQL backend (Row-Level Security) and a Next.js live alerts dashboard including GeoIP enrichment and AlienVault OTX intel.

As an AI Security Engineer intern, I built an LLM guardrail pipeline that blocked prompt injection, command execution, and PII leakage, aligned to HIPAA security requirements, using Python, FastAPI, Guardrails, and Azure OpenAI. I also reduced adversarial input risk by an estimated 40% by integrating the VirusTotal API into secure browser and sandbox agents, and I bring this same production mindset to every system I ship.