cybersecurity
Posts
- A technical analysis of the agentic runtime that dominated LinkedIn and GitHub with 150,000 stars and three name changes.
- Is hiding your model weights behind a REST API actually secure? A technical deep dive into how AugSteal, linear algebra, and side-channels are dismantling MLaaS.
Background
-
Paper: A Novel Approach for Detecting Forged SMTP Headers using Deep Learning and Synthetic Data Generation
Accepted in the Main Track of SBSeg 2025, Brazil’s premier cybersecurity symposium, achieving the maximum distinction (Four Seals of Approval).
- Excellence: One of the select few papers to receive all four approval seals, validating its technical depth and innovation.
- Innovation: Proposed a new architecture combining Deep Learning and Synthetic Data to detect sophisticated phishing attacks.
- Recognition: Derived from my undergraduate thesis and presented in person in Foz do Iguaçu, representing CEFET/RJ.