atom is a novel intermediate representation for applications and a standalone tool that is powered by chen.
ML research on software vulnerabilities
Advanced vulnerability management and analysis through data science techniques
ShieldPatch — Predict. Protect. Prevail. AI-driven vulnerability detection and automated patch prioritization platform using ML risk scoring, live threat intelligence (NVD, EPSS, ExploitDB), sandboxed patch testing, and an AI chatbot.
Cybersecurity threat intelligence for AI agents — CVE search, EPSS exploit prediction, CISA KEV, IP reputation, threat feed. MCP + x402.
Built an AI-powered vulnerability intelligence platform using NVD, CISA KEV, EPSS, BGE embeddings, XGBoost, FAISS, SHAP, and Streamlit to predict exploit likelihood for 357K+ CVEs, achieving ROC-AUC 0.99 and PR-AUC 0.87.
Exploit-likelihood modeling validated strictly forward in time: point-in-time features, forward-chaining CV, and calibration-first metrics — built to avoid the temporal leakage that inflates naive models.
Empirical Security — independent third-party profile of a public API surface, by API Evangelist. Empirical Security builds data-driven models that predict which vulnerabilities will actually be exploited, so security teams can prioritize remediation by real-world risk instead of raw CVSS severity.
To associate your repository with the exploit-prediction topic, visit your repo's landing page and select "manage topics."