A collection of research projects and explorations I've undertaken.
Built an auditing framework to detect and classify dark patterns in ride-hailing applications. This work involves systematic analysis of UI/UX patterns that manipulate user decisions, developed during my research internship at IIT BHU under Prof. N.S. Rajput.
A replication study examining the feasibility of data poisoning attacks on large language models under limited computational resources. Investigates how small, carefully crafted poisoned datasets can compromise model behavior even when the attacker has constrained hardware.
Research into internal auditing methods for machine learning models with full access to model weights, architectures, and training data. Focuses on inspecting decision boundaries, gradient-based analysis, and identifying potential failure modes before deployment.
Exploring quantum key distribution and secure communication channels. Studying how quantum mechanical principles can be leveraged to build provably secure communication protocols that are resistant to both classical and quantum computational attacks.
Studying the robustness of vision-language agents against targeted adversarial perturbations. Investigating how carefully crafted inputs can mislead multimodal systems that combine visual perception with language understanding, and exploring potential defense mechanisms.