Sneha Patil

Engineered an advanced Machine Learning Gradient Boosting model to categorize U.S. regions by wildfire risk levels, leveraging spatio-temporal and environmental variables. Architected a dynamic Streamlit application for real-time wildfire risk predictions, integrating historical fire datasets with geospatial features for enhanced decision-making. Fine-tuned model performance through advanced techniques to mitigate class imbalance, delivering predictive insights to optimize resource allocation and support proactive disaster management strategies.

Built an interactive BB84 protocol based QKD simulator using Qiskit and Streamlit to visualize quantum-secure key exchange. Simulated quantum eavesdropping to demonstrate intrusion detection via measurement-induced error rates and enabled message encryption using quantum-generated keys to showcase real-world QKD applications. Presented statistical diagnostics of key agreement rates to validate the integrity of quantum-secure exchanges.

Conducted research on content-based video retrieval, developing methods to search videos using visual inputs instead of semantic metadata. Engineered a multi-stage pipeline integrating video framing, face detection, face recognition, and SURF-based feature extraction for robust visual matching. Evaluated the system’s ability to retrieve relevant videos through facial feature analysis, demonstrating improved retrieval accuracy for large and diverse video datasets.