Projects
Focus
A few things I keep coming back to.
AI security
Finding and fixing prompt injection and system-prompt extraction in production LLM apps.
Reverse engineering
Mapping how AI tools work under the hood: system prompts, agent loops, model routing.
Open source
Building tools in public and maintaining them long after launch.
Web
React, Next.js, and TypeScript for the things around it.
Elsewhere
The best way to reach me is by email at lucasvalbuena@pm.me. You can also find more of my code on GitHub and more of my work on X and LinkedIn.