Research & CortexRouter
My research combines manual source and binary auditing with deterministic static analysis and CortexRouter—our in-house offline zero-day vulnerability discovery engine.
CortexRouter Zero 1.0 — 119B MoE Zero-Day Discovery Model
Trained on Mistral Small 4 (119B total parameters, 128-expert hybrid Mixture-of-Experts architecture, 256K context window), CortexRouter Zero 1.0 performs deep multi-step reasoning across full source repositories, ASTs, and protocol state machines to uncover authentication bypasses, cryptographic flaws, and memory-safety vulnerabilities.
CortexRouter Judge v1 (Apache-2.0 · Open Weights)
Open-weights Qwen3-8B LoRA vulnerability-evidence verification model producing structured 9-field triage decisions (Confirmed / Patched / False Positive) with 80/80 (100%) accuracy on our published 20-CVE clean holdout benchmark.
1. Authentication, Parsers & State Machines
Auditing trust boundaries in web backends, reverse proxies, authentication flows, and serialization parsers where subtle state desynchronization leads to privilege escalation or remote code execution.
2. CortexRouter Model Training & Variant Analysis
Training and scaling CortexRouter locally to recognize complex zero-day vulnerability patterns across large open-source ecosystems and firmware distributions without leaking sensitive code to external APIs.
3. Memory-Safe Auditing Tooling
Building minimal, zero-dependency verification tools in Rust and Go designed for reproducibility, low false-positive rates, and strict isolation.
Coordinated Advisories & Post-Patch Case Studies
Zero-day findings discovered with CortexRouter are reported free of charge to maintainers under our 90-day disclosure policy, with post-patch technical advisories published once users have had time to update.