Florian Martins

Software Engineer specializing in AI, Cyber Security & LLMOps/DevSecOps.

My focus lies at the intersection of Artificial Intelligence and Cybersecurity: building powerful AI systems while ensuring they remain resilient, reliable, and secure against modern threats.

Specializing in autonomous AI agent architectures and RAG (Retrieval-Augmented Generation) systems, I focus on LLM governance, structural guardrails, and mitigating AI-specific vulnerabilities (OWASP LLM Top 10, indirect prompt injections, data leaks).

Creator of open-source security & AI tools (Explore my projects here: github.com/FlorianMartins) :

• Regent & LLM-Guardrail-Proxy: Governance frameworks, security proxies, and mandate engines for securing autonomous AI agents and intercepting threats in real-time.

• LLM Security Lab & RAG-Eval-Pipeline: Automated red-team/pentesting suites and evaluation pipelines designed to rigorously test AI assistants and retrieval systems.

• Hivey Code & HiveyAI-Sidebar: Hybrid multi-provider LLM orchestration engines featuring local RAG (Transformers.js), Model Context Protocol (MCP) tool-calling loops, and reversible data pseudonymization.

• CloudGuard-IaC: Lightweight static analyzers and custom HCL2 parsers to secure Infrastructure-as-Code and Dockerfiles before deployment.

My mission: Enabling companies to safely deploy the power of LLMs into production (Cloud or local models via vLLM/Ollama) without compromising security, data privacy, or sovereignty.