The challenge

Important signals about cyber incidents are scattered across very different sources. News websites, government advisories, social platforms, Telegram channels, and dark-web forums each reveal only part of the picture—and reviewing them manually makes it easy to miss emerging threats.

What we built

Our five-person team built a real-time threat-intelligence platform that brings those signals together for monitoring cyber incidents across India. It continuously collects information from more than five source types and turns fragmented, unstructured data into a single stream that analysts can investigate.

I worked on the automated ingestion pipelines and enrichment workflow. Machine-learning models classify the sentiment and relevance of incoming content, while an LLM-assisted layer extracts and infers attacker tactics, techniques, and procedures (TTPs). The result is more useful than a collection of links: it gives analysts structured context about what may be happening and how an attacker is operating.

Recognition

The project won Smart India Hackathon 2024 under the National Technical Research Organisation’s cybersecurity domain. We presented the final solution at IIT Kharagpur in December 2024.

Built with: Python, web scraping, AWS, MongoDB, machine learning, and LLMs.