open-source
From LLMs to Secure Agents
I am building From LLMs to Secure Agents, a deep, visual, source-grounded guide to understanding complete agentic AI systems and learning how to secure them across 209 ordered units and two learning passes.
Hello, I'm Renato
Security Engineer.
Security engineer focused on secure agentic AI systems, with experience across systems security, AI-driven threat detection, and privacy-preserving authorization.
Open to opportunities
Overview
Security research and engineering for agentic AI systems.
I am a Security Engineer working across Security Research, Systems Security, AI for Cybersecurity, and Agentic AI Security. My current focus is understanding how agentic systems can be designed, evaluated, and secured across their tools, permissions, context, memory, execution, and human-control boundaries.
I am currently working with the ETH Agentic Systems Lab, focusing on agentic AI systems and their security, safety, reliability, robustness, evaluation, and real-world deployment.
Previously, at Huawei Research's Data and Privacy Protection Lab, I worked on privacy-enhancing technologies and privacy-preserving authorization. My thesis specified, implemented, and evaluated policy-based anonymous tokens with a post-quantum security target, aiming to verify scoped authorization facts while minimizing unnecessary identity disclosure and persistent tracking.
Selected projects
Selected projects across agent systems and security research.
open-source
VitaeContext
I co-created VitaeContext (formerly AgentKit SEO), an MIT-licensed open-source framework and MCP server that turns verified career context into scoped instructions for AI agents.
ctf
AI Agents - Reply Code Challenge 2026
I built a ReAct-based multi-agent system for the Reply Code Challenge 2026, finishing 57th globally out of 2000 teams.
Get in touch
Want to discuss agentic AI systems, security research, or an opportunity?
Always open to relocation for new experiences.