During the IASc–INSA–NASI Summer Research Fellowship Programme (SRFP) in June–July 2025, I worked on a research project focused on detecting Advanced Persistent Threats (APTs) in Industrial Internet of Things (IIoT) environments. This experience marked a significant phase in my academic development, as it enabled me to work closely on a real-world security problem that combines systems, networks, and machine learning. The core objective of my project was to design a near real-time APT detection framework suitable for industrial environments, where attacks are often stealthy, long-lasting, and difficult to identify using traditional methods. Instead of relying only on network traffic, the work leveraged system-level provenance data, which records detailed interactions between processes, files, and network sockets. These interactions were modeled as graphs to capture system behavior more accurately.

To extract meaningful patterns from these graphs, I used Node2Vec and Word2Vec to generate embeddings that represent system behavior in a compact form. Since security datasets are highly imbalanced, with very few attack samples compared to benign activity, Borderline-SMOTE was applied to improve learning around critical decision boundaries. The detection model was built using FastKAN, a lightweight and efficient neural architecture designed to achieve high recall with low inference time. The framework was evaluated on the CIC-APT-IIoT2024 dataset, where it demonstrated strong performance in terms of recall, F1-score, and inference speed, making it suitable for near real-time IIoT security monitoring.

Presenting this work at ICDCN 2026 was a meaningful and rewarding experience. Preparing for the presentation helped me reflect deeply on the motivation behind the work, the design choices, and the broader implications of the results. During the presentation, I received thoughtful questions and feedback from researchers working in distributed systems, networking, and cybersecurity. These discussions helped me better understand how my work fits into the larger research landscape and encouraged me to think about future improvements and extensions.

The conference also played a significant role in my personal and professional growth. Traveling solo and managing everything independently — from planning logistics to presenting my research — was a first-time experience for me, and one I truly enjoyed. Every part of the journey was new, and navigating it on my own built a sense of confidence and self-reliance. I had the opportunity to interact with researchers from different parts of the world, exchange ideas, and learn about ongoing research across institutions and countries. Everyone was welcoming, and engaging in conversations where we shared our research interests and experiences was truly enlightening. These interactions expanded my academic network and helped me feel more comfortable and confident discussing my work in an international setting. Observing keynote talks, technical sessions, and panel discussions further exposed me to a variety of research styles and perspectives.

Beyond the technical sessions, the breaks, reception, banquet, and cultural ritual offered space for relaxed conversations. Many of these discussions went beyond research — touching on culture, career journeys, and shared experiences as students and professionals. I had the chance to speak with senior researchers who appreciated my work and encouraged me to pursue it further. Their words of support meant a lot and made the experience even more memorable. On the last day of the conference, we began with a traditional deer-calling ritual, where participants offered special biscuits to Nara’s sacred deer — a charming and symbolic experience that reflected the host city’s cultural heritage.

I am deeply thankful to the ACM-W Scholarship program for supporting my participation in ICDCN 2026. The scholarship made it possible for me to attend the conference, present my research, and fully engage with the academic community.