Manish Tripathy usualdork

🛡️ Professional Contributions

🎯 OWASP Open Source Intelligence Standard

Aug 2026 – Present

Leading the OWASP Open Source Intelligence Standard (OSINT Standard) project, driving the vision, roadmap, and community collaboration. Developing an open standard for OSINT methodologies and best practices for the global cybersecurity community.

🎯 OWASP Maritime Security

Sep 2025 – Present

Contributing to the OWASP Maritime Security Top 10 project by developing vulnerability analysis frameworks, risk scoring methodologies, and threat intelligence resources to strengthen maritime cybersecurity guidance and support global security standards through open-source collaboration.

🔍 IEEE Conference Reviewer

Apr 2025 – Present

Reviewing research papers in Artificial Intelligence and Machine Learning for the IEEE INDISCON conference, contributing to the peer-review process and helping maintain the quality and integrity of published research.

📚 Research Publications & Technical Papers

🔬 Featured Research Papers

🔍 An Enhanced Framework for Surveillance: Integrating OSINT, SOCMINT, AI-Driven Analysis & Digital HUMINT Against Terrorism and Exploitation Networks

Publisher: IEEE

This paper proposes an advanced surveillance framework combining OSINT, SOCMINT, AI-driven analysis, and digital HUMINT for law enforcement.

💣 Unified Cyber Attack Vectors: OSINT, Reconnaissance, and Social Engineering

Publisher: Taylor & Francis

This paper explores how combining OSINT, reconnaissance, and social engineering enhances cyber attacks. Experiments show this integrated approach boosts phishing success by 135% and speeds up intelligence gathering by 40%.

🎭 Deceptive Digital Identities: When Social Media Profiles Are Engineered to Deceive: How Sophisticated Actors Architect Their Online Presence and a Counter-OSINT Framework to Detect Them

Publisher: Zenodo

This paper formalizes Content Asymmetry as a deception signal through the Content Asymmetry Index (CAI), applies graph-theoretic attack surface reduction to social identity, and presents a five-phase Counter-OSINT Reconnaissance framework. Grounded in an eight-year longitudinal study of 100,000+ profiles, it introduces a formal persona engineering taxonomy and a three-tier confidence stratification framework for deception-resistant profiling.

⚔️ Strategic Information Confrontation Through Persistent Digital Surveillance

Publisher: IEEE

Published at ICADS-26 (co-hosted with the Indian Army Air Defence College🎖️and IEEE), this paper introduces the Strategic Persistent Intelligence Confrontation (SPIC) framework: a military-grade, intelligence-warfare architecture.

🎯 Attention-Enhanced MTCNN for Strategic Face Detection

Publisher: IEEE

This paper presents a hardening of face detection for environments where cooperation, clarity, and control cannot be assumed. Rather than refining MTCNN at the surface level, this work restructures the entire detection cascade to operate under surveillance-grade conditions.

📖 Additional Research & Technical Contributions

🧠 Zero-Shot Learning for Face Recognition in Cyber Security

Publisher: Springer

The paper proposes a Zero-Shot Learning (ZSL) approach for face recognition in cybersecurity, enabling accurate recognition of unseen classes with limited data. Using semantic attributes from LFW, the ZSL model outperforms traditional methods.

🤖 Deep Learning Based Face Recognition with Modified MTCNN and FaceNet

Publisher: Taylor & Francis

The paper proposes an advanced face recognition system combining Modified MTCNN and FaceNet. Tested on three benchmark datasets: LFW, Political Faces, and Celebrity Faces, achieving 94.4% accuracy on the LFW dataset.

💰 AI-Driven Security Analytics for Zero-Footprint Due Diligence

Publisher: ICEVB-2025 Conference Proceedings

This research integrates AI, OSINT, and cybersecurity analytics to enable secure, automated profiling of rural entrepreneurs with no digital footprint.

🛡️ VANGUARD: A Cognitive Purple Agent Framework for Autonomous Adversarial Simulation and Real-Time SIEM Validation

Publisher: Zenodo

This paper introduces VANGUARD, a Cognitive Purple Agent fusing an LLM-driven ReAct Red Team engine with real-time Elasticsearch SIEM validation. Demonstrates 0.0% SOC detection rates against autonomously synthesized kill-chains, with immediate autonomous KQL rule deployment to close the gaps.

🧬 The Cognitive Fingerprint: LLM-Resistant Cross-Domain Threat Actor Attribution via Temporal-Syntactic Fusion

Publisher: TechRxiv (Preprint)

This paper proposes a novel attribution framework that bypasses the "LLM Wall" — the point where semantic models collapse against adversarially obfuscated text — by replacing Semantic Analysis with Behavioral Signal Processing. Introduces L-TDoA (Logical Time-Difference-of-Arrival) for measuring cross-platform cognitive latency, and SynGNN, a content-blind Graph Neural Network leveraging deep syntactic embeddings. SynGNN recovers 78.6% of identities under full structural stripping where BERT-based models degrade to 12.4%, achieving a simulated composite precision of 91.7% in controlled adversarial settings.