My research published in IRJMETS explored adversarial AI risks in cybersecurity. I developed BERT-based detection pipelines and lightweight, explainable defenses for email classifiers, with the reported F1 improving by 86.5%.
I built SHIELDMAIL to detect and defend against adversarial attacks on enterprise email classification models. It includes custom attack modules, a modular defense engine, and an offline visualization setup.
For my DDoS Detection System project, I designed a flow-based system that passively captures network traffic and detects anomalies using behavioral features, rule-based detection, and an Isolation Forest model.
At Digital Task Force, I assisted with cyber crime investigations, digital evidence collection, and forensic analysis. I also triaged security alerts and contributed research to threat intelligence reports.
