Research & Scientific Contributions
My research focuses on the intersection of Artificial Intelligence, Large Language Models, Natural Language Processing, and Cybersecurity, with particular emphasis on transforming unstructured information into actionable intelligence.
As a Scientist at the AIT Austrian Institute of Technology and a PhD Researcher at TU Wien, I investigate how AI-driven approaches can improve cyber threat intelligence analysis, automate complex analytical tasks, and support more reliable decision-making.
My work combines scientific evaluation with practical system development, exploring the capabilities and limitations of emerging AI technologies in real-world applications.
Research Areas
Large Language Models & Retrieval-Augmented Generation
Evaluating Large Language Models (LLMs), Retrieval-Augmented Generation (RAG), GraphRAG, hybrid retrieval architectures, and AI agents for complex information retrieval and question-answering tasks.
My research examines answer quality, factual grounding, hallucination, retrieval accuracy, computational efficiency, and the ability of AI systems to reason over structured and unstructured knowledge.
Cyber Threat Intelligence & Situational Awareness
Developing AI-powered approaches for extracting, processing, enriching, and analyzing Cyber Threat Intelligence (CTI) and Open-Source Intelligence (OSINT).
This includes named entity recognition, information extraction, text classification, knowledge graphs, and automated threat intelligence analysis.
MITRE ATT&CK & Threat Classification
Investigating automated identification and classification of adversarial Tactics, Techniques, and Procedures (TTPs) from cybersecurity reports using LLMs, semantic embeddings, and few-shot learning.
This research includes TTPFShot, a retrieval-based few-shot learning approach designed to classify MITRE ATT&CK techniques with minimal labeled data.
Machine Learning & Intelligent Information Systems
Comparing traditional machine learning algorithms, transformer-based embeddings, and large language models across cybersecurity classification and information retrieval tasks.
A central objective is understanding when simpler, computationally efficient machine learning methods can achieve performance comparable to, or better than, significantly larger AI models.
Selected Scientific Publications
2026
Beyond RAG for Cyber Threat Intelligence: A Systematic Evaluation of Graph-Based and Agentic Retrieval
2026 | arXiv Preprint
A systematic evaluation of vector-based, graph-based, agentic, and hybrid RAG architectures for cyber threat intelligence. The study investigates how knowledge graphs and structured reasoning improve complex question answering over threat intelligence reports.
Cybersecurity Text Classification: Challenging the Perceived Superiority of LLMs Over Conventional Machine Learning
IEEE International Conference on Big Data 2025 | Published 2026
An empirical comparison of traditional machine learning models, sentence embeddings, and open-source LLMs for multilingual cybersecurity text classification. The study demonstrates that efficient embedding-based approaches can achieve competitive performance at substantially lower computational cost.
2025
TTP Classification with Minimal Labeled Data: A Retrieval-Based Few-Shot Learning Approach
20th International Conference on Availability, Reliability and Security (ARES 2025) | Springer
Introducing TTPFShot, a retrieval-based few-shot learning framework for identifying adversarial techniques from cybersecurity reports using a minimal amount of labeled training data.
Enhancing Cyber Situational Awareness with AI: A Novel Pipeline Approach for Threat Intelligence Analysis and Enrichment
ARES 2025 International Workshops | Springer
Presenting an AI-driven pipeline that combines entity recognition, information extraction, classification, summarization, and knowledge representation to support automated cyber threat intelligence processing.
Evaluation and Comparison of Open-Source LLMs Using Natural Language Generation Quality Metrics
IEEE International Conference on Big Data 2024 | Published 2025
A comprehensive benchmarking study of open-source large language models using established natural language generation metrics and evaluation methods relevant to Retrieval-Augmented Generation.
Additional Collaborative Research
My scientific interests extend beyond cybersecurity into applied machine learning, predictive analytics, and intelligent decision-support systems.
Data-Driven Predictive Analytics for Dynamic Aviation Systems: Optimising Fleet Maintenance and Flight Operations Through Machine Learning
Future Internet, 2025
Research on machine learning methods for predictive aircraft maintenance and flight delay forecasting.
Machine Learning for an Enhanced Credit Risk Analysis: A Comparative Study of Loan Approval Prediction Models Integrating Mental Health Data
Machine Learning and Knowledge Extraction, 2024
An investigation of supervised machine learning approaches for credit risk assessment and loan approval prediction.
Current Research & Future Directions
My ongoing research explores the development of increasingly intelligent and reliable AI systems for cybersecurity.
Areas of particular interest include:
- Agentic AI for Cybersecurity: AI agents capable of interacting with external tools, structured knowledge, and security intelligence sources.
- Graph-Based Threat Intelligence: Combining knowledge graphs, semantic search, and LLM reasoning to analyze relationships between threat actors, vulnerabilities, malware, and attack techniques.
- APT Attribution: Machine learning and language model approaches for identifying potential threat actors from cyber threat intelligence reports.
- AI-Assisted Security Analysis: Context-aware systems that connect infrastructure information, vulnerability intelligence, and adversarial techniques to support security decision-making.
My long-term research objective is to bridge the gap between scientific advances in artificial intelligence and their practical application in cybersecurity.
Academic Profiles
Explore my scientific publications, research contributions, and academic activities through the following profiles.
DBLP Computer Science Bibliography
Research Collaboration
I welcome scientific collaborations, interdisciplinary research opportunities, and discussions on applying artificial intelligence to challenging problems in cybersecurity and intelligent information systems.
Interested in collaborating?
