I founded Badcompany, a project that aims to completely eliminate the security exposure of Agents to prompt injection attacks.I co-founded mesh., a company providing infrastructure for young builders focused on cutting-edge technology and creating real value. My interests are building solutions that actually solve problems.

AI security, Agents, ML engineering and architecture development, builder talent development

Badcompany: Building verifiably secure agents with cryptographically bounding tools mesh. : mesh. is a hackerlab, where we seek, invest in, and elevate builder talent.

Engineering Physics student with a background in Physics, Software Development and Biology, complemented by practical experience through internships and research. Participated in national and international science olympiads. Learning and doing projects in cybersecurity and ML.

Cybersecurity AI agents ML AI Interpretability Physics

I'm currently working on 3 projects: - AI agents cybersecurity: Developing a security protocol for safety-first agent deployment. - Chaos in LLMs: Using the tools from non-linear dynamics and chaos theory to understand the internal dynamics of LLMs. Interpretability, Data analysis, modelling. - Physics simulations: For granular material inmersed in fluids. Statistical physics and thermodynamics.

I like to train ML models and write fast and secure code.

electronic music

generative text2lego, AI red teaming agent, Sentinel: architecture for secure AI agent systems