Hardening & Zero-Trust RAG System
Thesis project: a FastAPI and LLM-based AI security assistant designed around Zero-Trust principles.
Role
Solo design, development, Docker containerization, and deployment (thesis project).
About the Project
My Computer Engineering thesis project. An AI-powered security assistant built around system hardening and Zero-Trust security: a FastAPI backend, Docker containerization, observability via Prometheus and Grafana, and an LLM/RAG layer that generates hardening recommendations. I carried the same observability tooling (Prometheus/Grafana) I use in my SRE role directly into this project. Not currently live.
Stack
RAG systems can be black boxes that look like they're working while quietly returning wrong answers. I brought the same Prometheus/Grafana discipline I use for alerting in my SRE role into this project, so I could see exactly where a pipeline run breaks down — retrieval, prompt construction, or the LLM response — not just whether it 'worked'.