Research Domain Interests

• Healthcare AI & Medical Imaging
  

  My research in healthcare AI focuses on advancing medical imaging and predictive analytics for improved clinical outcomes. I have worked on developing active learning frameworks, AutoML systems, and data-driven approaches that enhance image interpretation and diagnostic accuracy. Beyond imaging, my contributions extend to disease prediction, such as childhood and pregnancy lead poisoning and diabetic retinopathy, where deep learning models provide early detection and decision support. I have also explored conversational AI, designing personalized healthcare chatbots powered by federated learning and GPT-based architectures to improve accessibility and patient engagement in clinical contexts.
• Federated Learning & Privacy-Preserving Machine Learning
  

  A central theme of my work is federated learning (FL) and privacy-aware distributed AI. I have designed collaborative FL frameworks that integrate differential privacy to protect sensitive medical data, as well as hierarchical and clustered FL strategies (e.g., SCALE, SAFARI) to improve scalability in heterogeneous environments. My research also applies FL in practical security contexts, such as contraband detection from airport baggage X-rays, demonstrating the value of decentralized intelligence in real-world high-stakes applications. Additionally, I have investigated the integration of blockchain and generative AI with FL to ensure transparency, trust, and adaptability in privacy-critical domains.
• Social Media, NLP & Generative AI
  

  I actively explore natural language processing and generative AI to better understand and safeguard online social environments. My work includes tracking sentiment dynamics, detecting emotions, and building datasets for low-resource languages like Bangla to advance emotion and hate-speech detection. I have addressed identity deception and gender identification in online spaces, proposing solutions to enhance integrity in social networks. At the same time, I leverage LLMs (GPT-4, RAG, multimodal summarization, empathetic AI) for applications such as content filtering, personalized recommendations, and predictive modeling of user evolution. Together, these contributions aim to balance personalization, safety, and fairness in digital platforms.
• Cybersecurity & Quantum AI for Nuclear Systems
  

  A growing strand of my research applies AI to critical infrastructure security, particularly nuclear power plants. I have contributed to frameworks that integrate quantum cryptography, quantum-inspired federated learning, and large language models to enhance resilience against cyber threats such as DDoS attacks and anomaly intrusions. These works emphasize privacy-aware detection, predictive maintenance using digital twins, and contextually informed monitoring for safe operations. By combining cybersecurity, AI, and quantum technologies, my research advances the development of proactive, intelligent defense mechanisms for next-generation energy systems where reliability and safety are paramount.

UPDATES