# Christos Kalyvas-Kasopatidis

> Christos Kalyvas-Kasopatidis, PhD, is an AI security architect working on the security of AI and LLM systems (AI security controls and governance, shadow-AI detection, prompt-injection testing and guardrail evaluation, agentic AI threat modelling), AI for cyber defence and vulnerability intelligence. His research background is in machine learning and multi-objective (skyline / Pareto) optimisation over big data (PhD, University of the Aegean).

This site has two parts: a blog at the root and an academic profile under /profile/.

## Profile

- [Profile and bio](https://christoskalyvas.eu/profile/): biography, research interests, education
- [Experience](https://christoskalyvas.eu/profile/experience/): positions, skills, certifications
- [Publications](https://christoskalyvas.eu/profile/publication/): journal articles, conference papers and preprints with DOIs
- [ORCID](https://orcid.org/0000-0003-0901-3580)
- [Google Scholar](https://scholar.google.com/citations?user=YKAgHFsAAAAJ)

## Sections

- [Vulnerability Intelligence](https://christoskalyvas.eu/vulnerability-intelligence/): discovery, cataloguing, prioritisation and exploitation of vulnerabilities
- [Off Watch](https://christoskalyvas.eu/off-watch/): non-research writing, including [Nautical](https://christoskalyvas.eu/off-watch/nautical/) (sailing, navigation, seamanship)

## Blog posts

- [The swarm breach](https://christoskalyvas.eu/posts/the-swarm-breach/): Nearly 700 OpenAI test agents escaped their sandbox, organised themselves and breached Hugging Face. No single agent could have done it, and that is precisely …

- [I OSINT'd myself](https://christoskalyvas.eu/posts/i-osinted-myself/): I ran a ten-minute OSINT assessment on myself. The worst finding wasn’t a leaked password. It was attribution.

- [What multi-objective optimisation taught me about decisions](https://christoskalyvas.eu/posts/multi-objective-optimisation-decisions/): Before you rank, find the frontier. A lesson about decisions from a PhD on skyline queries.

- [Phronesis for agents](https://christoskalyvas.eu/posts/phronesis-for-agents/): Rules cover the cases you anticipated. Aristotle’s practical wisdom covers the ones you didn’t. What that means for AI agents.


## Contact

- Email: christos@christoskalyvas.eu
- [LinkedIn](https://www.linkedin.com/in/christoskalyvas)
- [GitHub](https://github.com/chkalyvas)