Profile banner
Thomas Roccia

Thomas Roccia

Senior security researcher, founder of SecurityBreak, threat intelligence author

thomas-roccia
France
Australia
Joined July 2026
Favicon imagesecuritybreak.ioFavicon imagex.comFavicon imageunprotect.itFavicon imageamazon.comFavicon imagelinkedin.com

Network

3.3K connections

Summary

Seasoned threat researcher with deep hands-on expertise in malware analysis, reverse engineering, incident response, and threat intelligence. He has performed global incident response, malware hunting and collaborated with law enforcement on major outbreaks. mcafee+2
Builder of open-source tools and frameworks for adversarial AI and prompt/agent monitoring — notably NOVA (prompt pattern matching), NovaHunting, Unprotect, and multiple AI-assisted tooling experiments that integrate RAG, MCP and agent workflows for CTI and malware triage. securitybreak+2
Public-facing educator and trainer: mentors on OpenClassrooms' EU Cybersecurity programs and delivers hands-on trainings and workshops (Black Hat, SANS, Botconf and other conferences) focused on practical malware analysis, threat hunting and AI for threat intelligence. openclassrooms+2
Author and communicator who publishes long-form technical writing and an award-winning book (Visual Threat Intelligence) to make threat intelligence accessible through visual explanations, practical tooling and reproducible research. securitybreak+2
Focused on the intersection of AI and security: pivoting recent research and offerings toward AI Threat Intelligence, adversarial AI, and agent monitoring — translating long experience in traditional CTI and malware research into the AI security domain. securitybreak+2

Work

Education

Projects

Writing

Inside the AI x Security Arsenal I’ve Built

May 1, 2025

A long-form blog/Medium post cataloguing 20+ AI tools and experiments for malware triage, CTI, RAG systems, MCP clients, NOVA, and other AI-for-security projects with notebooks and code.

Favicon imageblog.securitybreak.io

Introducing NOVA. The Prompt Pattern Matching

March 1, 2025

Explains the NOVA framework for prompt pattern matching (keywords, semantic similarity, and LLM-based evaluation) to detect malicious or adversarial prompts in AI systems and exported logs.

Favicon imageblog.securitybreak.io

Applying LLMs to Threat Intelligence (Talk / Video)

January 1, 2024

Presentation on applying LLMs and AI tools to threat intelligence workflows and automation (conference talk / recorded session).

Favicon imageyoutube.com

Visual Threat Intelligence: An Illustrated Guide For Threat Researchers

January 1, 2023

A concise, illustrated book covering threat intelligence fundamentals, TTPs, MITRE ATT&CK, practical tools (YARA, Sigma, MSTICpy), and case studies of notable cyberattacks; winner of the Foreword INDIES Bronze award (Science & Technology).

Favicon imagebook.securitybreak.io

Hobbies

3D printing (designed and printed a SecurityBreak mascot named 'Neon'). nitter