For my thesis, A Multimodal AI-Based System for Student Engagement Detection in Online Learning Environments, I architected and deployed a full-stack system for real-time engagement detection. As lead developer across the stack, I worked with React, Django REST API, PostgreSQL, and WebSockets; a live deployment achieved 193 ms mean AI inference.
I trained a lightweight LSTM on emotional and rPPG features from the DAiSEE dataset, reaching 0.34 recall on the “Disengaged” class. I also built a retrieval-augmented chatbot for uploaded PDFs that provides page-level citations and declines out-of-scope questions.
At Smart Sytems Laboratory, I design, research, and develop smart systems focused on IoT, cyberphysical systems, AI and machine learning, security and privacy, and smart applications. My projects also include deploying a corn leaf disease classifier with FastAPI, Docker, and CI/CD.

