Pierre-François Maillard
Security Analyst
Dr. Pierre-François Maillard is a security analyst specialized in AI and low-level security. He earned his engineering/master's degree in cybersecurity at ESIEA (France), and his PhD from TCG CREST (India), with a focus on adversarial machine learning. Over the past years, he has spoken at academic and industry conferences worldwide (c0c0n, ICSRS, etc.) and authored various articles in journals (MISC, JCVHT, etc.). He currently works at eShard as a Security Analyst and Customer Success, where he focuses on applying time travel debugging and AI to reverse engineering.
Reconstructing Telegram: Time Travel Debugging Meets AI on Android
Telegram is one of the world's most security-sensitive Android applications, which makes it an ideal case study. In this talk, we show how the combination of full-system Time Travel Debugging (TTD) and AI reconstructed Telegram's processing from a real 7 billion-instruction execution trace, without source code, hooks, or prior instrumentation.
We explain how full-system TTD records an application's complete execution, including its interactions with Android internals, turning transient runtime behavior into persistent data that both humans and AI can explore. We then show how this static representation lets AI overcome the context-window limitations that normally constrain debugger-driven analysis. Through the Telegram investigation, attendees will see how this approach exposes transitions, framework interactions, and low-level system behavior that conventional debugging often misses. Following that, they will see how this method extends to QEMU-based full-system emulation on iOS.